Photo Mosaicing. Sam Baird Ben Felsted Zach Gildersleeve David Koop. April 30, 2007

Size: px
Start display at page:

Download "Photo Mosaicing. Sam Baird Ben Felsted Zach Gildersleeve David Koop. April 30, 2007"

Transcription

1 Photo Mosaicing Sam Baird Ben Felsted Zach Gildersleeve David Koop April 30, Introduction While diverse work has been invested in mosaicing images together, the common goal is usually to combine smaller images to generate a single larger image. In applications such as aerial photography or medical imaging, the smaller images typically originate from the same dataset, and thus each subimage is a small part of a unified whole. Mosaicing can also be thought, more artistically, as a patchwork quilt. Instead of combining similar images to form a larger image, we use a collection of diverse images to compose a single image so that the new image looks from afar to be a single image, but upon closer inspection reveals the individual images. This set of images is typically referred to as a photomosaic. Many different approaches have been taken, but few have closely examined the strengths and weaknesses of each algorithm. We have researched and implemented many different techniques in order to compare. We start out with a simple localized mean squared error technique. We build upon that by introducing antipole clustering to speed up the search. We also analyze boundary matching and discrete cosine transform in choosing a best match. Finally, our work explores global optimization as a technique for creating the mosaics, expanding the local optimization algorithm to include an iterative simulated annealing that converges on an optimal solution given the subimage dataset. 2 Related Work The idea behind photo mosaicing dates back to 1973, when Harmon use block portraits to study human perception and pattern recogntion. He presented a very low resolution image of Abraham Lincoln, where each pixel was a large tile. When the subject stood far away, they could recognize the face. Then in 1976, Dali created a painting similar to Harmon s, called Lincoln in Dalivision, but he included a woman in the center. When looking close-up, you could only recognize the woman. When viewing from far away, all you saw was Lincoln. The first software implementation of photomosaicing was started by Silver while he was at MIT. While earlier attempts used Photoshop to replace a pixel with an image, Silvers replaced a tile (grid of pixels) with an image. He matched the image according to color and shape. In 1998, Hunt took this idea a step farther and implemented photo mosaicing using different size tiles. Klein also furthered Silver s idea by creating a video mosaic. Instead of an image being created from a set of smaller images, a video was created from a set of smaller videos. This adds a level of complexity because time has to now be accounted for in matching. Because 1

2 consistency in the small videos is desired, it is hard to find good matches. To alleviate this problem, they correct the color to match the current frame. Another algorithm was proposed in 1998 that was based off Silver s idea. After choosing the best match for the subimages, each image is altered according to color to best match the target tile. It calculates the desired average color and creates a correction function such that the final average color in the subimage is the same as the average color in the target tile. Most recently, researchers have tackled the problem of panoramic mosaic, where the final image is a set of photographs of the same scene, stitched together from different perspectives. However, the goal is not to notice the stiches and thus make a coherent final image. Typically, you preprocess the image database and extract particular information about each image. For example, you can resize the image for fast comparisons, you can compute their histograms, or you can build a hierarchy to speed up the search process. De Blasi was able to speed up the search process using the Antipole strategy. Using an Antipole Tree Data Structure, you embed a large set of images into a metric space (X, d). Each image is a point in X and distance d is defined as the mean squared error in RGB values. Images can then be grouped into clusters so that far elements lie in different clusters. You define endpoints and then partition the points according to the proximity to the two endpoints. This is repeated again recursively on the two clusters. This gives you a binary tree in which to search. 3 Algorithm Framework While the concept of a photomosaic is fairly straightforward, the exact specification of it is less refined. We are trying to accurately replicate an image I using a set of given images S. In order to make this more precise, we need to define a metric that captures this accuracy. Given the image I, we define a function f : I S such that f(x, y) = s(g(x, y)) S. Thus, f defines 1. the image s S from which the pixel is to be taken 2. the location in s, g(x, y), from which the pixel is to be taken Given this function, we can define a metric based on the difference between I(x, y) and f(x, y). Of course, given this definition, we can simply construct an f that takes the pixel closest to I(x, y) from the set S. This is counterintuitive to the mosaic idea where f(x, y) is taken from the same image as its neighbors unless it lies along a boundary. Thus, we also need to impose a locality constraint on f to enforce this idea. Let B denote a set of intrapixels {(x+1/2, +1/2)}. f(x, y) and f(x ± 1, y ± 1) must map to the same image s unless (x ± 1/2, y ± 1/2) is in B. Often B is a gridded structure, but it may be different shapes as well. Most photomosaics are constructed using a local optimization, meaning that for each contiguous section of our decomposition of I, we find the best s S. However, this idea can be extended to instead consider a global optimization. For this problem, we constrain our selections from S to disallow repetition. Thus, picking the best local match may cause a poor global match. We can still score each candidate image against the region to be painted, but instead of simply taking the candidate with the highest score, we select a candidate with this global view. 2

3 Specifically, let I be decomposed into regions R(I) = {R 1,..., R m }. We wish to determine a mapping m : R(I) S such that m match(r i, m(r i )) i=1 is maximized. As stated, this is a local optimization problem. However, suppose we restrict this mapping so that m(r i ) = m(r j ) = i = j. Then, we have a global optimization problem. 3.1 Greedy Algorithm We first implemented a simple greedy algorithm. We choose a region R(I) at random and select an image from S that best matches the region. The error between R(I) and the image in S is defined by a mean squared error on the RGB values of the pixels. When duplication of images in S is allowed, an optimal arrangement of images is easily found. However, duplication is usually not desired. To remedy this, we take two different approaches. In one method, we penalize images that have already been used, thus enforcing a wider variety of images. In our other method, we don t allow duplication. Finding a global optimal answer becomes very difficult when this uniqueness constraint is enforced. We explore simulated annealing later in this paper in order to itteratively approach the optimal answer. 3.2 Antipole Clustering To choose the best match, we have to search through each image in set S. This needs to be done for every region R(I). This search can be cumbersome considering how many images there are in the database (we have 1500) and how many regions we compute (typically around 5,000-10,000). To speed up this process, we use an antipole clustering approach. By defining a metric space, we can represent an image in S as a point in space. We can define the distance between two points as the mean squared error between the two images. Each image in S is a point, and we can cluster the points based on distance. First, we find the farthest two points from each other. Then we go through each point and cluster it to the closest endpoint. This creates two clusters. We recurse on each cluster until we have only one point for each cluster. We now have a binary tree of clusters. When we process a particular region R(I) of our final image I, we search for the closest match in S. However, we can use our tree structure to find the correct image in logarthimic time instead of linear time. We find which cluster the region is closer to, and recurse down that branch, until we ve found the best match. Because this created a lot of overhead, the benefit of using an antipole structure was not evident with a small image I. However, for large images, the clustering approach was four times faster than the linear approach when a I was a 250x250 image. 4 Boundary Matching In the area of texture synthesis, the major criteria for a good synthesis is that the tesselation boundaries are invisible. For photomosaics, we will get a much better approximation to the original image if we succeed in matching these boundaries. In some cases, this is desirable to 3

4 Figure 1: An Example of Antipole Clustering obtain a better approximation, but often the distinct boundaries are desired to maintain the photomosaic structure. Thus, for adjoining candidate images, we can calculate their boundary matching score by examining the pixels near the corresponding edge. Intuitively, we wish this boundary to be as smooth as possible. This can be obtained by examining the gradient across the edge. With boundaries, our optimization problem is now m (1 α) match(r i, m(r i )) + α i=1 R j B(R i ) boundary(m(r i ), m(r j ), e(r i, R j )) where α represents the weighting of boundary matching. Our work is not increased too much. For each image, we need to compute its match with each region, and for each pair of images, we need to compute its boundary match along each possible edge. We could also match the boundaries by doing a mean squared error on the boundary edges. We add the boundary error to the error between the region and the image from S. However, by adding another error metric, we sacrifice the selection quality. We have a smoother image, but our image quality decreases because it selects suboptimal matches. We have implemented this algorithm in our work. We notice that as we weight the boundary criteria more, the recognition of the image decreases. 5 Discrete Cosine Transformations While the described averaging method is a fairly straightforward means of creating a photomosaic, it is by no means the only method. In theory, any algorithm that provides a means for ranking the similarity between two images can be used to drive the photomosaic. We present an exploration using the Discrete Cosine Transformation (DCT) to rank the subimages and provide a means for comparison to each cell of the master image. The DCT is known best for its use in lossy JPEG compression due to its strong energy compaction property. Compared to the similar Discrete Fourier Transformation, the DCT concentrates most of its energy in lower frequencies. These lower frequencies represent the general outline of objects in an image; JPEG compression is achieved by clamping the DCT transformed image, thus limiting the detail in solid, filled areas. 4

5 There are multiple definitions of the DCT, we will use the the most common DCT-II: D k = N 1 n=0 [ ( π x n cos n + 1 ) ] k N 2 k = 0,..., N 1 To transform a 2D image using the DCT, significant acceleration can be found by using square images. If I is a square image with side n, the DCT of I can be found by I DCT = D I D where the nxn DCT transformation matrix D need only be computed once. 5.1 DCT Photomosaic Algorithm The basic algorithm to use the DCT as matching function within a photomosaic application is as follows. First, we calculate the DCT of each subimage as above. This is done using a precomputed DCT transformation matrix of mxm, where m typically is set to 128 pixels, which we can use throughout the implementation. Due to the high energy compactness of the DCT, it is necessary to only compare the highest n DCT coefficients of each subimage. The value of n is introduced as a parameter, and provides results that quickly converge on the best match for a set of subimages numbering in the hundreds. Beyond the highest n coefficients, no change of R I is possible, due to the effective quantization of a finite number of subimages, but additional coefficients could be introduced as a means to alter R I. The master image I is reduced to its individual regions R(I) = {R 1,..., R m }, as in the initial implementation, but these regions are of arbitrary size (typically smaller than mxm) which does not correspond to the mxm DCT transformation matrix. It is worth noting that D nxn I nxn D nxn D mxm I mxm D mxm therefore we interpolate the cell B to the size of the DCT transformation matrix using a interpolation matrix H m. B m = H m B s B m = D m B m D m B n = B m (1 : n) The result, in DCT space, can be compared against the set of subimages using mean squared error to find the best fit, using either a local or global optimization as discussed. The comparison can be done against the full m 2 values in the DCT images, or clamped to the highest n values as in JPEG compression. If the described photomosaic matching algorithm is conducted in DCT space, matching subimages are located at the low frequency edges of the master image, at the expense of detail in the filled areas. Thus, this algorithm focuses on the edges of an image. However, since we are now matching the lower frequencies of the subimage to the region R i, the result better provides subregion matching, reducing pixelation. 5

6 5.2 DCT Results and Discussion Using the DCT to drive the photomosaic algorithm produces decidedly different results than the initial localized mean square error method. In that method, each subregion within a block is matched according to the average color value. With the DCT method, we are comparing only against frequency. Thus, we get slightly different results. Figure 2: Goal Image (a) (b) 3x3 Averaging DCT Matching Figure 3: Comparison of Photomosaic Results As we can see in Figure 3, in particular the hair region of the goal image Figure 2, the DCT matching method matches the blocks at a subblock level better than the average matching method. Figure 3 is only matching frequency, and not average color, and therefore some of the large solid areas, such as the hat and shirt, are matched to a subimage that shares a similar frequency signature, but is divergent in terms of overall color average. This is to be expected, as this information is essentially lost when we only compare the top n values of our DCT signature. An important observation is that for discrete images, high frequencies at the global scale translate to low frequencies at the local scale; that is, if we interpolate a small local region 6

7 to a much larger size, any high frequencies are expanded out and essentially become low frequencies. As the energy compactness property of the DCT maintains a large proportion of these low frequencies, much of the original image is preserved and, in the application of photomosaic, used to match the sub images. 5.3 Extension of the DCT to Image Searching The Discrete Cosine Tranform, like its Fourier cousin, finds many uses. In addition to lossy compression, visualized using a photomosaic algorithm, the DCT can be used for other shape (low frequency) matching and searching utilities. Bracamonte et al, among others, have noticed that the broad DCT algorithm outlined above can be used for quick image based searching. Such searching can be done in the DCT domain, which for web outlets where JPEG images are the norm, this can account for considerable acceleration. While the details of such an algorithm fall outside the scope of this discussion, a brief examination of image based searching is possible. If we use a block size equal to the goal image we are essentially comparing against our database of subimages the image that best matches the overall low frequency signature of the goal image. This can provide some interesting results, although a complete robust search algorithm would most likely be interested in additional parameters such as color. In Figure 4 we see the fullsized subimage that best matches the highest 256 DCT coefficients of the same goal image, out of 1500 images. This relatively simple concept can be very powerful in its utility. (a) (b) Goal Image DCT Matching Figure 4: DCT Based Image Searching 6 Simulated Annealing The problem of finding a global minimum in error involves a search over a large state space. Each state is a permutation of the ordering of tiles that are to be matched with the goal image. 7

8 In the experiments we performed, the number of permutations was the following: ( ) Number of tiles 1501! Number of blocks! = Number of blocks ( )! Simulated Annealing is a generic probabilistic meta-algorithm for finding a global minimum in a large state space. It iteratively converges to a solution of an NP-problem using a polynomial time algorithm at each iteration. Name of Simulated Annealing (SA) comes from annealing in metallurgy, in which material is heated and cooled in a controlled manner. The goal of this is to increase the crystal size of the material and decrease defects. The addition of heat allows molecules to take a highenergy random walk to become unstuck from local minima of energy configurations. As the temperature cools, the molecules take smaller steps and the total energy eventually converges to a new minimum. SA is closely related to the physical process, except it is applied to the global optimization problem. Each step of SA replaces the current solution with a neighboring solution. The probability of choosing a particular solution depends on the energy (or error) of the current and next solution and on the current temperature, T. The probability distribution is chosen such that when T is large, the probability is almost uniform, and when T is small, the solution converges in the Cauchy sense. Additionally, when T becomes small, the probability is close to one if transitioning to the next solution causes a decrease in energy, and close to zero otherwise. In the context of Photo Mosiacing, we make the following defining assumption: 1. Each state is a permutation of the tiles. We denote the current and next state as e and e. 2. Neighboring states differ by a swap of tiles 3. The energy of state is proportional to the sum of squared errors The probability of a state transition due to Kirpatrick et. al: Pr(e, e, T ) = α exp e e T The formula comes from Metropolis-Hastings algorithm used to generate samples from the Maxwell-Boltzmann distribution. Figure 5 shows the convergence of total energy for a given number of iterations. The final image converged to a solution that had small error on average, but a large error for at least one of the tiles. Using the greedy method resulted in errors that were relatively small on average. 7 Final Discussion At first glance, the photomosaic concept is a purely artistic endeavor. However, powerful concepts fundamental to image processing, signal compression, and global optimization can be visualized via a photomosaic. In the 3x3 algorithm, a simple photomosaic framework was 8

9 3.15 x 10 4 Simulated Annealing: Total energy for a given number of steps 3.1 Total energy Number of iterations Figure 5: Trend of Simulated Annealing Optimization established to provide a testbed for technique comparison. Antipole clustering introduced a binary tree acceleration structure to photomosaicing that substantially increased the speed required to produce an image. For large I, the same image could be determined in roughly a fourth of the time. With this acceleration, it was possible to further refine the photomosaic algorithm, by introducing boundary matching to find subimages R i that matched both the 3x3 block, as well as matched the nearest n blocks to create as smooth a gradient as possible given the set of subimages. To continue this thought, the photomosaic algorithm was further extended to explore global optimization using an iterative simulated annealing optimization. As an open solution problem, using simulated annealing introduces additional questions such as the number of required iterations before an adequate convergence of the final image. Given a discrete dataset of subimages, an ideal solution is theoretically possible given enough processing time, but at what point remains an area for further research. Finally, we explored matching R i with the Discrete Cosine Transform in the frequency domain rather than matching a nearest-neighbor average in the image domain. While this direction also remains an avenue for further work, matching the frequency rather than color allows for more complex subimage matching, as well as low frequency searching and retrieval. References [1] BATTIATO, S., DI BLASI, G., FARINELLA, G. M., AND GALLO, G. Digital mosaic frameworks, [2] JAVIER BRACAMONTE AND ANSORGE AND PELLANDINI AND PIERRE-ANDRÉ FARINE. Efficient Compressed Domain Target Image Search and Retrieval. In CIVR 05: Proceedings of the 4th International Conference on Image and Video Retrieval pp [3] DI BLASI, G., AND PETRALIA, M. Fast photomosaic, [4] FINKELSTEIN, A., AND RANGE, M. Image mosaics. In EP 98/RIDT 98: Proceedings of the 7th International Conference on Electronic Publishing (London, UK, 1998), Springer- Verlag, pp

10 [5] KIM, J., AND PELLACINI, F. Jigsaw image mosaic. In SIGGRAPH 2002 Conference Proceedings (2002), vol. 38, pp [6] KLEIN, A. W., GRANT, T., FINKELSTEIN, A., AND COHEN, M. F. Video mosaics. In NPAR 2002: Second International Symposium on Non Photorealistic Rendering (2002), pp [7] SILVERS, R. Photomosaics. Henry Holt and Co., Inc., New York, NY, USA, [8] TRAN, N. Generating photomosaics: an empirical study. In SAC 99: Proceedings of the 1999 ACM symposium on Applied computing (New York, NY, USA, 1999), ACM Press, pp [9] ZHANG, Y. On the use of cbir in image mosaic generation,

Fast Techniques for Mosaic Rendering

Fast Techniques for Mosaic Rendering Fast Techniques for Mosaic Rendering Gianpiero Di Blasi, Giovanni Gallo, Petralia Maria D.M.I. - Università di Catania Summary Mosaics and computer Related works Fast techniques for mosaic rendering Artificial

More information

A Hillclimbing Approach to Image Mosaics

A Hillclimbing Approach to Image Mosaics A Hillclimbing Approach to Image Mosaics Chris Allen Faculty Sponsor: Kenny Hunt, Department of Computer Science ABSTRACT This paper presents a hillclimbing approach to image mosaic creation. Our approach

More information

Jigsaw Image Mosaics (JIM)

Jigsaw Image Mosaics (JIM) Jigsaw Image Mosaics (JIM) Based on the paper Jigsaw Image Mosaics by Junhwan Kim and Fabio Pellacini, SIGGRAPH 2002 April 1, 2004 Presentation by Kaleigh Smith Outline Description of JIM, artistic inspiration

More information

(Refer Slide Time 00:17) Welcome to the course on Digital Image Processing. (Refer Slide Time 00:22)

(Refer Slide Time 00:17) Welcome to the course on Digital Image Processing. (Refer Slide Time 00:22) Digital Image Processing Prof. P. K. Biswas Department of Electronics and Electrical Communications Engineering Indian Institute of Technology, Kharagpur Module Number 01 Lecture Number 02 Application

More information

Robust PDF Table Locator

Robust PDF Table Locator Robust PDF Table Locator December 17, 2016 1 Introduction Data scientists rely on an abundance of tabular data stored in easy-to-machine-read formats like.csv files. Unfortunately, most government records

More information

Computer Vision 2. SS 18 Dr. Benjamin Guthier Professur für Bildverarbeitung. Computer Vision 2 Dr. Benjamin Guthier

Computer Vision 2. SS 18 Dr. Benjamin Guthier Professur für Bildverarbeitung. Computer Vision 2 Dr. Benjamin Guthier Computer Vision 2 SS 18 Dr. Benjamin Guthier Professur für Bildverarbeitung Computer Vision 2 Dr. Benjamin Guthier 1. IMAGE PROCESSING Computer Vision 2 Dr. Benjamin Guthier Content of this Chapter Non-linear

More information

Rendering Technique for Colored Paper Mosaic

Rendering Technique for Colored Paper Mosaic Rendering Technique for Colored Paper Mosaic Youngsup Park, Sanghyun Seo, YongJae Gi, Hanna Song, and Kyunghyun Yoon CG Lab., CS&E, ChungAng University, 221, HeokSuk-dong, DongJak-gu, Seoul, Korea {cookie,shseo,yj1023,comely1004,khyoon}@cglab.cse.cau.ac.kr

More information

Jigsaw Image Mosaics. Kim and Pellacini, Cornell Presented by Kacper Wysocki

Jigsaw Image Mosaics. Kim and Pellacini, Cornell Presented by Kacper Wysocki Jigsaw Image Mosaics Kim and Pellacini, Cornell 2004 Presented by Kacper Wysocki Introduction Fill arbitrary container compactly with arbitrarily shaped tiles of similar color Deform tiles slightly for

More information

Homework # 2 Due: October 6. Programming Multiprocessors: Parallelism, Communication, and Synchronization

Homework # 2 Due: October 6. Programming Multiprocessors: Parallelism, Communication, and Synchronization ECE669: Parallel Computer Architecture Fall 2 Handout #2 Homework # 2 Due: October 6 Programming Multiprocessors: Parallelism, Communication, and Synchronization 1 Introduction When developing multiprocessor

More information

Segmentation and Grouping

Segmentation and Grouping Segmentation and Grouping How and what do we see? Fundamental Problems ' Focus of attention, or grouping ' What subsets of pixels do we consider as possible objects? ' All connected subsets? ' Representation

More information

Adaptive-Mesh-Refinement Pattern

Adaptive-Mesh-Refinement Pattern Adaptive-Mesh-Refinement Pattern I. Problem Data-parallelism is exposed on a geometric mesh structure (either irregular or regular), where each point iteratively communicates with nearby neighboring points

More information

Motivation. Intensity Levels

Motivation. Intensity Levels Motivation Image Intensity and Point Operations Dr. Edmund Lam Department of Electrical and Electronic Engineering The University of Hong ong A digital image is a matrix of numbers, each corresponding

More information

Non-Photorealistic Experimentation Jhon Adams

Non-Photorealistic Experimentation Jhon Adams Non-Photorealistic Experimentation Jhon Adams Danny Coretti Abstract Photo-realistic rendering techniques provide an excellent method for integrating stylized rendering into an otherwise dominated field

More information

Understanding Clustering Supervising the unsupervised

Understanding Clustering Supervising the unsupervised Understanding Clustering Supervising the unsupervised Janu Verma IBM T.J. Watson Research Center, New York http://jverma.github.io/ jverma@us.ibm.com @januverma Clustering Grouping together similar data

More information

EE 701 ROBOT VISION. Segmentation

EE 701 ROBOT VISION. Segmentation EE 701 ROBOT VISION Regions and Image Segmentation Histogram-based Segmentation Automatic Thresholding K-means Clustering Spatial Coherence Merging and Splitting Graph Theoretic Segmentation Region Growing

More information

Lecture 4. Convexity Robust cost functions Optimizing non-convex functions. 3B1B Optimization Michaelmas 2017 A. Zisserman

Lecture 4. Convexity Robust cost functions Optimizing non-convex functions. 3B1B Optimization Michaelmas 2017 A. Zisserman Lecture 4 3B1B Optimization Michaelmas 2017 A. Zisserman Convexity Robust cost functions Optimizing non-convex functions grid search branch and bound simulated annealing evolutionary optimization The Optimization

More information

Image Analysis - Lecture 5

Image Analysis - Lecture 5 Texture Segmentation Clustering Review Image Analysis - Lecture 5 Texture and Segmentation Magnus Oskarsson Lecture 5 Texture Segmentation Clustering Review Contents Texture Textons Filter Banks Gabor

More information

Fast Fuzzy Clustering of Infrared Images. 2. brfcm

Fast Fuzzy Clustering of Infrared Images. 2. brfcm Fast Fuzzy Clustering of Infrared Images Steven Eschrich, Jingwei Ke, Lawrence O. Hall and Dmitry B. Goldgof Department of Computer Science and Engineering, ENB 118 University of South Florida 4202 E.

More information

The Plan: Basic statistics: Random and pseudorandom numbers and their generation: Chapter 16.

The Plan: Basic statistics: Random and pseudorandom numbers and their generation: Chapter 16. Scientific Computing with Case Studies SIAM Press, 29 http://www.cs.umd.edu/users/oleary/sccswebpage Lecture Notes for Unit IV Monte Carlo Computations Dianne P. O Leary c 28 What is a Monte-Carlo method?

More information

An Approach for Reduction of Rain Streaks from a Single Image

An Approach for Reduction of Rain Streaks from a Single Image An Approach for Reduction of Rain Streaks from a Single Image Vijayakumar Majjagi 1, Netravati U M 2 1 4 th Semester, M. Tech, Digital Electronics, Department of Electronics and Communication G M Institute

More information

Particle Swarm Optimization applied to Pattern Recognition

Particle Swarm Optimization applied to Pattern Recognition Particle Swarm Optimization applied to Pattern Recognition by Abel Mengistu Advisor: Dr. Raheel Ahmad CS Senior Research 2011 Manchester College May, 2011-1 - Table of Contents Introduction... - 3 - Objectives...

More information

IMAGE COMPRESSION USING FOURIER TRANSFORMS

IMAGE COMPRESSION USING FOURIER TRANSFORMS IMAGE COMPRESSION USING FOURIER TRANSFORMS Kevin Cherry May 2, 2008 Math 4325 Compression is a technique for storing files in less space than would normally be required. This in general, has two major

More information

Scientific Computing with Case Studies SIAM Press, Lecture Notes for Unit IV Monte Carlo

Scientific Computing with Case Studies SIAM Press, Lecture Notes for Unit IV Monte Carlo Scientific Computing with Case Studies SIAM Press, 2009 http://www.cs.umd.edu/users/oleary/sccswebpage Lecture Notes for Unit IV Monte Carlo Computations Dianne P. O Leary c 2008 1 What is a Monte-Carlo

More information

Mesh Simplification. Mesh Simplification. Mesh Simplification Goals. Mesh Simplification Motivation. Vertex Clustering. Mesh Simplification Overview

Mesh Simplification. Mesh Simplification. Mesh Simplification Goals. Mesh Simplification Motivation. Vertex Clustering. Mesh Simplification Overview Mesh Simplification Mesh Simplification Adam Finkelstein Princeton University COS 56, Fall 008 Slides from: Funkhouser Division, Viewpoint, Cohen Mesh Simplification Motivation Interactive visualization

More information

Simple Silhouettes for Complex Surfaces

Simple Silhouettes for Complex Surfaces Eurographics Symposium on Geometry Processing(2003) L. Kobbelt, P. Schröder, H. Hoppe (Editors) Simple Silhouettes for Complex Surfaces D. Kirsanov, P. V. Sander, and S. J. Gortler Harvard University Abstract

More information

Chapter 1. Introduction

Chapter 1. Introduction Chapter 1 Introduction A Monte Carlo method is a compuational method that uses random numbers to compute (estimate) some quantity of interest. Very often the quantity we want to compute is the mean of

More information

Topic: Local Search: Max-Cut, Facility Location Date: 2/13/2007

Topic: Local Search: Max-Cut, Facility Location Date: 2/13/2007 CS880: Approximations Algorithms Scribe: Chi Man Liu Lecturer: Shuchi Chawla Topic: Local Search: Max-Cut, Facility Location Date: 2/3/2007 In previous lectures we saw how dynamic programming could be

More information

SYDE Winter 2011 Introduction to Pattern Recognition. Clustering

SYDE Winter 2011 Introduction to Pattern Recognition. Clustering SYDE 372 - Winter 2011 Introduction to Pattern Recognition Clustering Alexander Wong Department of Systems Design Engineering University of Waterloo Outline 1 2 3 4 5 All the approaches we have learned

More information

ADAPTIVE TEXTURE IMAGE RETRIEVAL IN TRANSFORM DOMAIN

ADAPTIVE TEXTURE IMAGE RETRIEVAL IN TRANSFORM DOMAIN THE SEVENTH INTERNATIONAL CONFERENCE ON CONTROL, AUTOMATION, ROBOTICS AND VISION (ICARCV 2002), DEC. 2-5, 2002, SINGAPORE. ADAPTIVE TEXTURE IMAGE RETRIEVAL IN TRANSFORM DOMAIN Bin Zhang, Catalin I Tomai,

More information

Digital Image Processing

Digital Image Processing Digital Image Processing Lecture # 4 Digital Image Fundamentals - II ALI JAVED Lecturer SOFTWARE ENGINEERING DEPARTMENT U.E.T TAXILA Email:: ali.javed@uettaxila.edu.pk Office Room #:: 7 Presentation Outline

More information

Evolutionary Computation Algorithms for Cryptanalysis: A Study

Evolutionary Computation Algorithms for Cryptanalysis: A Study Evolutionary Computation Algorithms for Cryptanalysis: A Study Poonam Garg Information Technology and Management Dept. Institute of Management Technology Ghaziabad, India pgarg@imt.edu Abstract The cryptanalysis

More information

CHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION

CHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION CHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION 6.1 INTRODUCTION Fuzzy logic based computational techniques are becoming increasingly important in the medical image analysis arena. The significant

More information

EE795: Computer Vision and Intelligent Systems

EE795: Computer Vision and Intelligent Systems EE795: Computer Vision and Intelligent Systems Spring 2012 TTh 17:30-18:45 FDH 204 Lecture 14 130307 http://www.ee.unlv.edu/~b1morris/ecg795/ 2 Outline Review Stereo Dense Motion Estimation Translational

More information

CSE 573: Artificial Intelligence Autumn 2010

CSE 573: Artificial Intelligence Autumn 2010 CSE 573: Artificial Intelligence Autumn 2010 Lecture 16: Machine Learning Topics 12/7/2010 Luke Zettlemoyer Most slides over the course adapted from Dan Klein. 1 Announcements Syllabus revised Machine

More information

Diffusion Wavelets for Natural Image Analysis

Diffusion Wavelets for Natural Image Analysis Diffusion Wavelets for Natural Image Analysis Tyrus Berry December 16, 2011 Contents 1 Project Description 2 2 Introduction to Diffusion Wavelets 2 2.1 Diffusion Multiresolution............................

More information

Operators-Based on Second Derivative double derivative Laplacian operator Laplacian Operator Laplacian Of Gaussian (LOG) Operator LOG

Operators-Based on Second Derivative double derivative Laplacian operator Laplacian Operator Laplacian Of Gaussian (LOG) Operator LOG Operators-Based on Second Derivative The principle of edge detection based on double derivative is to detect only those points as edge points which possess local maxima in the gradient values. Laplacian

More information

Locally Weighted Least Squares Regression for Image Denoising, Reconstruction and Up-sampling

Locally Weighted Least Squares Regression for Image Denoising, Reconstruction and Up-sampling Locally Weighted Least Squares Regression for Image Denoising, Reconstruction and Up-sampling Moritz Baecher May 15, 29 1 Introduction Edge-preserving smoothing and super-resolution are classic and important

More information

Robotics Programming Laboratory

Robotics Programming Laboratory Chair of Software Engineering Robotics Programming Laboratory Bertrand Meyer Jiwon Shin Lecture 8: Robot Perception Perception http://pascallin.ecs.soton.ac.uk/challenges/voc/databases.html#caltech car

More information

Image Mosaics with Irregular Tiling

Image Mosaics with Irregular Tiling Image Mosaics with Irregular Tiling Lina Zhang State Key Lab. of CAD&CG Zhejiang University Hangzhou, China zhanglina@zjucadcg.cn Jinhui Yu State Key Lab. of CAD&CG Zhejiang University Hangzhou, China

More information

Title: Adaptive Region Merging Segmentation of Airborne Imagery for Roof Condition Assessment. Abstract:

Title: Adaptive Region Merging Segmentation of Airborne Imagery for Roof Condition Assessment. Abstract: Title: Adaptive Region Merging Segmentation of Airborne Imagery for Roof Condition Assessment Abstract: In order to perform residential roof condition assessment with very-high-resolution airborne imagery,

More information

University of Cambridge Engineering Part IIB Module 4F12 - Computer Vision and Robotics Mobile Computer Vision

University of Cambridge Engineering Part IIB Module 4F12 - Computer Vision and Robotics Mobile Computer Vision report University of Cambridge Engineering Part IIB Module 4F12 - Computer Vision and Robotics Mobile Computer Vision Web Server master database User Interface Images + labels image feature algorithm Extract

More information

Multigrid Pattern. I. Problem. II. Driving Forces. III. Solution

Multigrid Pattern. I. Problem. II. Driving Forces. III. Solution Multigrid Pattern I. Problem Problem domain is decomposed into a set of geometric grids, where each element participates in a local computation followed by data exchanges with adjacent neighbors. The grids

More information

Tracking system. Danica Kragic. Object Recognition & Model Based Tracking

Tracking system. Danica Kragic. Object Recognition & Model Based Tracking Tracking system Object Recognition & Model Based Tracking Motivation Manipulating objects in domestic environments Localization / Navigation Object Recognition Servoing Tracking Grasping Pose estimation

More information

Classification: Feature Vectors

Classification: Feature Vectors Classification: Feature Vectors Hello, Do you want free printr cartriges? Why pay more when you can get them ABSOLUTELY FREE! Just # free YOUR_NAME MISSPELLED FROM_FRIEND... : : : : 2 0 2 0 PIXEL 7,12

More information

Clustering Color/Intensity. Group together pixels of similar color/intensity.

Clustering Color/Intensity. Group together pixels of similar color/intensity. Clustering Color/Intensity Group together pixels of similar color/intensity. Agglomerative Clustering Cluster = connected pixels with similar color. Optimal decomposition may be hard. For example, find

More information

CGAL. Mesh Simplification. (Slides from Tom Funkhouser, Adam Finkelstein)

CGAL. Mesh Simplification. (Slides from Tom Funkhouser, Adam Finkelstein) CGAL Mesh Simplification (Slides from Tom Funkhouser, Adam Finkelstein) Siddhartha Chaudhuri http://www.cse.iitb.ac.in/~cs749 In a nutshell Problem: Meshes have too many polygons for storage, rendering,

More information

Light Field Occlusion Removal

Light Field Occlusion Removal Light Field Occlusion Removal Shannon Kao Stanford University kaos@stanford.edu Figure 1: Occlusion removal pipeline. The input image (left) is part of a focal stack representing a light field. Each image

More information

CS4442/9542b Artificial Intelligence II prof. Olga Veksler

CS4442/9542b Artificial Intelligence II prof. Olga Veksler CS4442/9542b Artificial Intelligence II prof. Olga Veksler Lecture 2 Computer Vision Introduction, Filtering Some slides from: D. Jacobs, D. Lowe, S. Seitz, A.Efros, X. Li, R. Fergus, J. Hayes, S. Lazebnik,

More information

Lecture 27: Fast Laplacian Solvers

Lecture 27: Fast Laplacian Solvers Lecture 27: Fast Laplacian Solvers Scribed by Eric Lee, Eston Schweickart, Chengrun Yang November 21, 2017 1 How Fast Laplacian Solvers Work We want to solve Lx = b with L being a Laplacian matrix. Recall

More information

Geometric Registration for Deformable Shapes 3.3 Advanced Global Matching

Geometric Registration for Deformable Shapes 3.3 Advanced Global Matching Geometric Registration for Deformable Shapes 3.3 Advanced Global Matching Correlated Correspondences [ASP*04] A Complete Registration System [HAW*08] In this session Advanced Global Matching Some practical

More information

Compression of RADARSAT Data with Block Adaptive Wavelets Abstract: 1. Introduction

Compression of RADARSAT Data with Block Adaptive Wavelets Abstract: 1. Introduction Compression of RADARSAT Data with Block Adaptive Wavelets Ian Cumming and Jing Wang Department of Electrical and Computer Engineering The University of British Columbia 2356 Main Mall, Vancouver, BC, Canada

More information

ECE 533 Digital Image Processing- Fall Group Project Embedded Image coding using zero-trees of Wavelet Transform

ECE 533 Digital Image Processing- Fall Group Project Embedded Image coding using zero-trees of Wavelet Transform ECE 533 Digital Image Processing- Fall 2003 Group Project Embedded Image coding using zero-trees of Wavelet Transform Harish Rajagopal Brett Buehl 12/11/03 Contributions Tasks Harish Rajagopal (%) Brett

More information

Applications. Foreground / background segmentation Finding skin-colored regions. Finding the moving objects. Intelligent scissors

Applications. Foreground / background segmentation Finding skin-colored regions. Finding the moving objects. Intelligent scissors Segmentation I Goal Separate image into coherent regions Berkeley segmentation database: http://www.eecs.berkeley.edu/research/projects/cs/vision/grouping/segbench/ Slide by L. Lazebnik Applications Intelligent

More information

JPEG Compression. What is JPEG?

JPEG Compression. What is JPEG? JPEG Compression Michael W. Chou Scott Siegrist EEA Spring April, Professor Ingrid Verbauwhede What is JPEG? JPEG is short for the 'Joint Photographic Experts Group'. The JPEG standard is fairly complex

More information

Lecture 10: Semantic Segmentation and Clustering

Lecture 10: Semantic Segmentation and Clustering Lecture 10: Semantic Segmentation and Clustering Vineet Kosaraju, Davy Ragland, Adrien Truong, Effie Nehoran, Maneekwan Toyungyernsub Department of Computer Science Stanford University Stanford, CA 94305

More information

Module 1 Lecture Notes 2. Optimization Problem and Model Formulation

Module 1 Lecture Notes 2. Optimization Problem and Model Formulation Optimization Methods: Introduction and Basic concepts 1 Module 1 Lecture Notes 2 Optimization Problem and Model Formulation Introduction In the previous lecture we studied the evolution of optimization

More information

A System of Image Matching and 3D Reconstruction

A System of Image Matching and 3D Reconstruction A System of Image Matching and 3D Reconstruction CS231A Project Report 1. Introduction Xianfeng Rui Given thousands of unordered images of photos with a variety of scenes in your gallery, you will find

More information

Machine Learning A W 1sst KU. b) [1 P] Give an example for a probability distributions P (A, B, C) that disproves

Machine Learning A W 1sst KU. b) [1 P] Give an example for a probability distributions P (A, B, C) that disproves Machine Learning A 708.064 11W 1sst KU Exercises Problems marked with * are optional. 1 Conditional Independence I [2 P] a) [1 P] Give an example for a probability distribution P (A, B, C) that disproves

More information

Discrete geometry. Lecture 2. Alexander & Michael Bronstein tosca.cs.technion.ac.il/book

Discrete geometry. Lecture 2. Alexander & Michael Bronstein tosca.cs.technion.ac.il/book Discrete geometry Lecture 2 Alexander & Michael Bronstein tosca.cs.technion.ac.il/book Numerical geometry of non-rigid shapes Stanford University, Winter 2009 The world is continuous, but the mind is discrete

More information

Markov Random Fields and Gibbs Sampling for Image Denoising

Markov Random Fields and Gibbs Sampling for Image Denoising Markov Random Fields and Gibbs Sampling for Image Denoising Chang Yue Electrical Engineering Stanford University changyue@stanfoed.edu Abstract This project applies Gibbs Sampling based on different Markov

More information

1 Case study of SVM (Rob)

1 Case study of SVM (Rob) DRAFT a final version will be posted shortly COS 424: Interacting with Data Lecturer: Rob Schapire and David Blei Lecture # 8 Scribe: Indraneel Mukherjee March 1, 2007 In the previous lecture we saw how

More information

Learning and Inferring Depth from Monocular Images. Jiyan Pan April 1, 2009

Learning and Inferring Depth from Monocular Images. Jiyan Pan April 1, 2009 Learning and Inferring Depth from Monocular Images Jiyan Pan April 1, 2009 Traditional ways of inferring depth Binocular disparity Structure from motion Defocus Given a single monocular image, how to infer

More information

A Review of Image Compression Techniques

A Review of Image Compression Techniques A Review of Image Compression Techniques Rajesh, Gagan Kumar Computer Science and Engineering Department, MIET College, Mohri, Kurukshetra, Haryana, India Abstract: The demand for images, video sequences

More information

ISSN (ONLINE): , VOLUME-3, ISSUE-1,

ISSN (ONLINE): , VOLUME-3, ISSUE-1, PERFORMANCE ANALYSIS OF LOSSLESS COMPRESSION TECHNIQUES TO INVESTIGATE THE OPTIMUM IMAGE COMPRESSION TECHNIQUE Dr. S. Swapna Rani Associate Professor, ECE Department M.V.S.R Engineering College, Nadergul,

More information

Optimization Techniques for Design Space Exploration

Optimization Techniques for Design Space Exploration 0-0-7 Optimization Techniques for Design Space Exploration Zebo Peng Embedded Systems Laboratory (ESLAB) Linköping University Outline Optimization problems in ERT system design Heuristic techniques Simulated

More information

A Developer s Survey of Polygonal Simplification algorithms. CS 563 Advanced Topics in Computer Graphics Fan Wu Mar. 31, 2005

A Developer s Survey of Polygonal Simplification algorithms. CS 563 Advanced Topics in Computer Graphics Fan Wu Mar. 31, 2005 A Developer s Survey of Polygonal Simplification algorithms CS 563 Advanced Topics in Computer Graphics Fan Wu Mar. 31, 2005 Some questions to ask Why simplification? What are my models like? What matters

More information

Medical Image Segmentation Based on Mutual Information Maximization

Medical Image Segmentation Based on Mutual Information Maximization Medical Image Segmentation Based on Mutual Information Maximization J.Rigau, M.Feixas, M.Sbert, A.Bardera, and I.Boada Institut d Informatica i Aplicacions, Universitat de Girona, Spain {jaume.rigau,miquel.feixas,mateu.sbert,anton.bardera,imma.boada}@udg.es

More information

Final Review. Image Processing CSE 166 Lecture 18

Final Review. Image Processing CSE 166 Lecture 18 Final Review Image Processing CSE 166 Lecture 18 Topics covered Basis vectors Matrix based transforms Wavelet transform Image compression Image watermarking Morphological image processing Segmentation

More information

Automatic Generation of An Infinite Panorama

Automatic Generation of An Infinite Panorama Automatic Generation of An Infinite Panorama Lisa H. Chan Alexei A. Efros Carnegie Mellon University Original Image Scene Matches Output Image Figure 1: Given an input image, scene matching from a large

More information

CS 664 Segmentation. Daniel Huttenlocher

CS 664 Segmentation. Daniel Huttenlocher CS 664 Segmentation Daniel Huttenlocher Grouping Perceptual Organization Structural relationships between tokens Parallelism, symmetry, alignment Similarity of token properties Often strong psychophysical

More information

Optimizing the Deblocking Algorithm for. H.264 Decoder Implementation

Optimizing the Deblocking Algorithm for. H.264 Decoder Implementation Optimizing the Deblocking Algorithm for H.264 Decoder Implementation Ken Kin-Hung Lam Abstract In the emerging H.264 video coding standard, a deblocking/loop filter is required for improving the visual

More information

Introduction to Design Optimization: Search Methods

Introduction to Design Optimization: Search Methods Introduction to Design Optimization: Search Methods 1-D Optimization The Search We don t know the curve. Given α, we can calculate f(α). By inspecting some points, we try to find the approximated shape

More information

Evolved Multi-resolution Transforms for Optimized Image Compression and Reconstruction under Quantization

Evolved Multi-resolution Transforms for Optimized Image Compression and Reconstruction under Quantization Evolved Multi-resolution Transforms for Optimized Image Compression and Reconstruction under Quantization FRANK W. MOORE Mathematical Sciences Department University of Alaska Anchorage CAS 154, 3211 Providence

More information

Consistency and Set Intersection

Consistency and Set Intersection Consistency and Set Intersection Yuanlin Zhang and Roland H.C. Yap National University of Singapore 3 Science Drive 2, Singapore {zhangyl,ryap}@comp.nus.edu.sg Abstract We propose a new framework to study

More information

IMAGE SEGMENTATION. Václav Hlaváč

IMAGE SEGMENTATION. Václav Hlaváč IMAGE SEGMENTATION Václav Hlaváč Czech Technical University in Prague Faculty of Electrical Engineering, Department of Cybernetics Center for Machine Perception http://cmp.felk.cvut.cz/ hlavac, hlavac@fel.cvut.cz

More information

Motivation. Gray Levels

Motivation. Gray Levels Motivation Image Intensity and Point Operations Dr. Edmund Lam Department of Electrical and Electronic Engineering The University of Hong ong A digital image is a matrix of numbers, each corresponding

More information

Image-Space-Parallel Direct Volume Rendering on a Cluster of PCs

Image-Space-Parallel Direct Volume Rendering on a Cluster of PCs Image-Space-Parallel Direct Volume Rendering on a Cluster of PCs B. Barla Cambazoglu and Cevdet Aykanat Bilkent University, Department of Computer Engineering, 06800, Ankara, Turkey {berkant,aykanat}@cs.bilkent.edu.tr

More information

Classifying Images with Visual/Textual Cues. By Steven Kappes and Yan Cao

Classifying Images with Visual/Textual Cues. By Steven Kappes and Yan Cao Classifying Images with Visual/Textual Cues By Steven Kappes and Yan Cao Motivation Image search Building large sets of classified images Robotics Background Object recognition is unsolved Deformable shaped

More information

Multimedia Systems Image III (Image Compression, JPEG) Mahdi Amiri April 2011 Sharif University of Technology

Multimedia Systems Image III (Image Compression, JPEG) Mahdi Amiri April 2011 Sharif University of Technology Course Presentation Multimedia Systems Image III (Image Compression, JPEG) Mahdi Amiri April 2011 Sharif University of Technology Image Compression Basics Large amount of data in digital images File size

More information

Panoramic Image Stitching

Panoramic Image Stitching Mcgill University Panoramic Image Stitching by Kai Wang Pengbo Li A report submitted in fulfillment for the COMP 558 Final project in the Faculty of Computer Science April 2013 Mcgill University Abstract

More information

From Structure-from-Motion Point Clouds to Fast Location Recognition

From Structure-from-Motion Point Clouds to Fast Location Recognition From Structure-from-Motion Point Clouds to Fast Location Recognition Arnold Irschara1;2, Christopher Zach2, Jan-Michael Frahm2, Horst Bischof1 1Graz University of Technology firschara, bischofg@icg.tugraz.at

More information

Shweta Gandhi, Dr.D.M.Yadav JSPM S Bhivarabai sawant Institute of technology & research Electronics and telecom.dept, Wagholi, Pune

Shweta Gandhi, Dr.D.M.Yadav JSPM S Bhivarabai sawant Institute of technology & research Electronics and telecom.dept, Wagholi, Pune Face sketch photo synthesis Shweta Gandhi, Dr.D.M.Yadav JSPM S Bhivarabai sawant Institute of technology & research Electronics and telecom.dept, Wagholi, Pune Abstract Face sketch to photo synthesis has

More information

Simulating Human Performance on the Traveling Salesman Problem Abstract Introduction

Simulating Human Performance on the Traveling Salesman Problem Abstract Introduction Simulating Human Performance on the Traveling Salesman Problem Bradley J. Best (bjbest@cmu.edu), Herbert A. Simon (Herb_Simon@v.gp.cs.cmu.edu) Carnegie Mellon University Department of Psychology Pittsburgh,

More information

CS6670: Computer Vision

CS6670: Computer Vision CS6670: Computer Vision Noah Snavely Lecture 19: Graph Cuts source S sink T Readings Szeliski, Chapter 11.2 11.5 Stereo results with window search problems in areas of uniform texture Window-based matching

More information

Fractal Image Compression. Kyle Patel EENG510 Image Processing Final project

Fractal Image Compression. Kyle Patel EENG510 Image Processing Final project Fractal Image Compression Kyle Patel EENG510 Image Processing Final project Introduction Extension of Iterated Function Systems (IFS) for image compression Typically used for creating fractals Images tend

More information

Query by Fax for Content-Based Image Retrieval

Query by Fax for Content-Based Image Retrieval Query by Fax for Content-Based Image Retrieval Mohammad F. A. Fauzi and Paul H. Lewis Intelligence, Agents and Multimedia Group, Department of Electronics and Computer Science, University of Southampton,

More information

ENERGY-224 Reservoir Simulation Project Report. Ala Alzayer

ENERGY-224 Reservoir Simulation Project Report. Ala Alzayer ENERGY-224 Reservoir Simulation Project Report Ala Alzayer Autumn Quarter December 3, 2014 Contents 1 Objective 2 2 Governing Equations 2 3 Methodolgy 3 3.1 BlockMesh.........................................

More information

Combinatorial optimization and its applications in image Processing. Filip Malmberg

Combinatorial optimization and its applications in image Processing. Filip Malmberg Combinatorial optimization and its applications in image Processing Filip Malmberg Part 1: Optimization in image processing Optimization in image processing Many image processing problems can be formulated

More information

Image Compression. -The idea is to remove redundant data from the image (i.e., data which do not affect image quality significantly)

Image Compression. -The idea is to remove redundant data from the image (i.e., data which do not affect image quality significantly) Introduction Image Compression -The goal of image compression is the reduction of the amount of data required to represent a digital image. -The idea is to remove redundant data from the image (i.e., data

More information

Multi-Label Moves for Multi-Label Energies

Multi-Label Moves for Multi-Label Energies Multi-Label Moves for Multi-Label Energies Olga Veksler University of Western Ontario some work is joint with Olivier Juan, Xiaoqing Liu, Yu Liu Outline Review multi-label optimization with graph cuts

More information

Supervised vs. Unsupervised Learning

Supervised vs. Unsupervised Learning Clustering Supervised vs. Unsupervised Learning So far we have assumed that the training samples used to design the classifier were labeled by their class membership (supervised learning) We assume now

More information

DIGITAL IMAGE WATERMARKING BASED ON A RELATION BETWEEN SPATIAL AND FREQUENCY DOMAINS

DIGITAL IMAGE WATERMARKING BASED ON A RELATION BETWEEN SPATIAL AND FREQUENCY DOMAINS DIGITAL IMAGE WATERMARKING BASED ON A RELATION BETWEEN SPATIAL AND FREQUENCY DOMAINS Murat Furat Mustafa Oral e-mail: mfurat@cu.edu.tr e-mail: moral@mku.edu.tr Cukurova University, Faculty of Engineering,

More information

In this part of the course, I will discuss various approaches for generating VPLs where they are most needed for a given camera view.

In this part of the course, I will discuss various approaches for generating VPLs where they are most needed for a given camera view. In this part of the course, I will discuss various approaches for generating VPLs where they are most needed for a given camera view. Let me start by reviewing the classic many lights rendering algorithm,

More information

Floorplan and Power/Ground Network Co-Synthesis for Fast Design Convergence

Floorplan and Power/Ground Network Co-Synthesis for Fast Design Convergence Floorplan and Power/Ground Network Co-Synthesis for Fast Design Convergence Chen-Wei Liu 12 and Yao-Wen Chang 2 1 Synopsys Taiwan Limited 2 Department of Electrical Engineering National Taiwan University,

More information

DiFi: Distance Fields - Fast Computation Using Graphics Hardware

DiFi: Distance Fields - Fast Computation Using Graphics Hardware DiFi: Distance Fields - Fast Computation Using Graphics Hardware Avneesh Sud Dinesh Manocha UNC-Chapel Hill http://gamma.cs.unc.edu/difi Distance Fields Distance Function For a site a scalar function f:r

More information

The Immersed Interface Method

The Immersed Interface Method The Immersed Interface Method Numerical Solutions of PDEs Involving Interfaces and Irregular Domains Zhiiin Li Kazufumi Ito North Carolina State University Raleigh, North Carolina Society for Industrial

More information

Edge and corner detection

Edge and corner detection Edge and corner detection Prof. Stricker Doz. G. Bleser Computer Vision: Object and People Tracking Goals Where is the information in an image? How is an object characterized? How can I find measurements

More information

Computer Vision. Image Segmentation. 10. Segmentation. Computer Engineering, Sejong University. Dongil Han

Computer Vision. Image Segmentation. 10. Segmentation. Computer Engineering, Sejong University. Dongil Han Computer Vision 10. Segmentation Computer Engineering, Sejong University Dongil Han Image Segmentation Image segmentation Subdivides an image into its constituent regions or objects - After an image has

More information

6. Object Identification L AK S H M O U. E D U

6. Object Identification L AK S H M O U. E D U 6. Object Identification L AK S H M AN @ O U. E D U Objects Information extracted from spatial grids often need to be associated with objects not just an individual pixel Group of pixels that form a real-world

More information

ELEC Dr Reji Mathew Electrical Engineering UNSW

ELEC Dr Reji Mathew Electrical Engineering UNSW ELEC 4622 Dr Reji Mathew Electrical Engineering UNSW Review of Motion Modelling and Estimation Introduction to Motion Modelling & Estimation Forward Motion Backward Motion Block Motion Estimation Motion

More information