PERSONALIZATION OF MESSAGES

Transcription

1 PERSONALIZATION OF MESSAGES Arun Pandian 1, Balaji 2, Gowtham 3, Harinath 4, Hariharan 5 1,2,3,4 Student, Department of Computer Science and Engineering, TRP Engineering College,Tamilnadu, India 5 Associate Professor, Department of Computer Science and Engineering, TRP Engineering College,Tamilnadu, India Abstract- spam, defined as unsolicited bulk , continues to be a major problem in the Internet. With the spread of malware combined with the power of bot-nets, spammers are now able to launch large scale spam campaigns covering wide range of topics (e.g., pharmaceutical products, adult content, etc.) causing major traffic increase and leading to enormous economical loss. Filtering image spam is considered to be a challenging problem because spammers keep modifying the images being used in their campaigns by employing different obfuscation techniques. Therefore, preventing text recognition using Optical Character Recognition (OCR) tools and imposing additional challenges in filtering such type of spam. In this project, we propose an image spam filtering technique, called Image visual features analysis that makes use of low-level image features for image characterization. And implement Support vector machine to classify the spam images or not. Our experimental studies based on social network framework show that the SVM classifier outperforms with improved accuracy. Keywords - , Security, Spam, Spammers, Campaigns,Unsolicited, Internet, CBIR, Image Spam, OCR, Filtering, Visual Features, Personalization, Classification, Prioritization and Unwanted. I. INTRODUCTION Image processing is a method to convert an image into digital form and perform some operations on it, in order to get an enhanced image or to extract some useful information from it. It is a type of signal dispensation in which input is image, like video frame or photograph and output may be image or characteristics associated with that image. Usually Image Processing system includes treating images as two dimensional signals while applying already set signal processing methods to them. Image processing is among rapidly growing technologies today, with its applications in various aspects of a business. Image Processing forms core research area within engineering and computer science disciplines too. Importing the image with optical scanner or by digital photography. Analyzing and manipulating the image which includes data compression and image enhancement and spotting patterns that are not to human eyes like satellite photographs. Image processing systems are becoming popular due to easy availability of powerful personal computer, large size memory devices, graphics software etc. Output is the last stage in which result can be altered image or report that is based on image analysis. The format of an image spam is the same as standard format of image over the Internet like Subject, Message Body, and Receiver Address. Match the recognized text with database for eliminate trained text which is considered as spam words. The rest of the paper is organized as follows. Section 1 presented some introduction on Image processing.the various applications are discussed in section 2. Related work is presented in section 3 and proposed work is given in section 4. Section 5 presents conclusion and future work. II. APPLICATIONS OF IMAGE All Rights Reserved 474

2 Importing the image with optical scanner or by digital photography. Analyzing and manipulating the image which includes data compression and image enhancement and spotting patterns that are not to human eyes like satellite photographs. Output is the last stage in which result can be altered image or report that is based on image analysis. Remote sensing Medical imaging Forensic Studies Textiles Material science Military Film industry Document Processing The common steps in image processing are image storing, scanning, enhancing and interpretation. The following Figure 1.1 shows image processing. Figure 1. Image Processing In the above Figure, an image has been captured by a camera and has been sent to a digital system to remove all the other details, and just focus on the water drop by zooming it in such a way that the quality of the image remains the same. III. RELATED WORK In existing system proposed approach based on Base64 encoding of image files and n-gram technique for feature extraction. It transformed normal images into Base64 presentation, and then it used n-gram technique to extract the feature. Using SVM, spam images were detected from legitimate images. This approach shows time efficient performance. And proposed supervised detection method builds its training dataset based on two image features Colour and gradient orientation histograms and utilizes this data on probabilistic boosting tree (PBT) to distinguish spam images from ham images. Each node of PBT contains colour or gradient orientation histogram data of corresponding part of images inside training dataset. In the proposed detection method postulated that spammers use the same template to send a lot of spam images and they add random noises to an image template in order to bypass filters. Authors classify random noises into 17 categories and utilize three feature vectors in order to analyze them. By extracting these features from images, the system builds training dataset, compares new images with dataset and labels them as spam or ham images. Then propose fast and robust image spam detection method for dealing with image spam in s. Image Processing forms core research area within engineering and computer science disciplines too. They extract 9 features from images for feeding the maximum entropy model (i.e, logistic regression based on binary case) to detect spam. They also use Just in Time (JIT) feature extraction to speed All Rights Reserved 475

3 process of spam detection that dramatically reduces processing time. JIT is a feature extraction method, which only focuses and extracts features based on each image. Hough transform detection method is used in our system to reduce time. IV. PROPOSED WORK spam filtering represents a major approach to combat spam. The goal of spam filtering is to classify messages into ham or spam. Content-based techniques inspect the body of an searching for specific keyword(s) or features that are typically used by spammers or associated by certain spam campaign. body itself may be text, image, or both. Therefore, content-based filtering techniques usually deal with all these content types. Probabilistic image modeling is particularly suited to the task at hand since we want to classify a family of images as spam, while having observed only a few samples from the family. We have chosen Gaussian Mixture Models (GMM) as the starting point for our approach. Gaussian Mixture Models model an image as coherent regions in feature space. First, features are extracted for each pixel, projecting the image to a high-dimensional feature space. For each pixel, we extract a seven tuple feature vector: two parameters for pixel coordinates (x, y) (to include spatial information), three for color attributes in color space and two for texture attributes (anisotropy and contrast ).And we explore visual features like texture, shape and color and learn classifiers using these selected features. Notice that we extract global features, i.e. each feature represents a property of the entire spam image. The philosophy of OCR-based techniques is based on extracting the text embedded into attached images, then the same approaches used in spam filters to analyze s body text is used, which are keyword detection and text categorization techniques. The power of OCR-based techniques is determined by the OCR system itself. OCR errors is considered as one of the drawbacks of this kind of filters, especially when spammers obscure the content of the image by adding noise, dots, changing the background colours and rotating images, which affects the efficiency of OCR text extraction. This fact has led to other techniques based on low-level image features and a combination of OCR with low-level image features Figure 2.Architecture All Rights Reserved 476

4 STEPS INVOLVED ARE AS FOLLOWS: INTERFACE CREATION- creates the interface link mail application. IMAGE ACQUISITION- User can upload the images from their storage. FEATURES EXTRACTION- Extract the features such as color, shape and texture features. CLASSIFICATION OF IMAGES- Using SVM algorithm to classify the image features based on training features. OCR RECOGNITION- OCR tool is used to extract the text from images. IMAGE SPAM DETECTION - same as standard format of image over the Internet like Subject, Message Body, and Receiver Address. Figure 3.Index The Application proposed for the work is given in Figure 3. The proposed Personalization system will be able to prioritize the messages into three categories. They are Fun, Threat, Sensitive. The words which are fun,threat and Sensitive oriented are already collected in a database and they are used to check the words will be given in the subject. Once the match occurs then that mail will be prioritized under respective All Rights Reserved 477

5 . Figure 4.Inbox In the above screenshot the prioritization is clearly represented. Depending upon the word entered in the subject of the mail, the received mail is categorized. Also background colour is changed respective to the category. For Fun category its is green, for Threat category it is Orange and For Sensitive category it is dark Rose.In this way the messages are categorized. V. CONCLUSION AND FUTURE WORK CONCLUSION In this project, we present an image spam detection system. By examining the content of new images contained in incoming s and detecting images that are near duplicates of known spam images, our system can effectively detect image spam while maintaining a low false positive rate. Rather than using computationally expensive algorithms to detect new types of image spam designed to thwart conventional computer vision algorithms, our system uses efficient algorithms to target randomization methods used to generate large number of unique but visually similar image spam s from template images. Our system is designed to be integrated with existing anti-spam technologies to boost the detection rate of image spam. FUTURE WORK We can extend the work on new feature extraction units for image spam filters that can improve the performance of the categories in which our current system does not perform well. Furthermore, since image spam is constantly evolving, we believe it is a constant battle to find new features that can effectively defeat new image spam techniques REFERENCES 1. K. Albrecht, N. Burri, and R. Wattenhofer. Spamato-An Extendable Spam Filter System. 2 nd Conference on and Anti-Spam (CEAS), Stanford University, Palo Alto, California, USA, C. Burges. A Tutorial on Support Vector Machines for Pattern Recognition. Data Mining and Knowledge Discovery, 2(2): , C. Carson, M. Thomas, S. Belongie, J. Hellerstein, and J. Malik. Blobworld: A system for region-based image indexing and retrieval. Third International Conference on Visual Information Systems, pages , A. Dempster, N. Laird, and D. Rubin. Maximum likelihood from incomplete data via the em algorithm. Journal of the Royal Statistical Society, 39(1):1 38, All Rights Reserved 478

6 5. M. Dredze, R. Gevaryahu, and A. Elias-Bachrach. Learning Fast Classifiers for Image Spam. In proceedings of the Conference on and Anti-Spam (CEAS), 2007, pages , G. Fumera, I. Pillai, and F. Roli. Spam Filtering Based On The Analysis Of Text Information Embedded Into Images. The Journal of Machine Learning Research, 7: , J. Goldberger, S. Gordon, and H. Greenspan. An efficient image similarity measure based on approximations of KLdivergence between two gaussian mixtures. Computer Vision, Proceedings. Ninth IEEE International Conference on, pages , J. Goldberger, H. Greenspan, and S. Gordon. Unsupervised Image clustering using the Information Bottleneck method. Proc. DAGM, R. Haralick, I. Dinstein, and K. Shanmugam. Textural features for image classification. IEEE Transactions on Systems, Man, and Cybernetics, 3: , T. Joachims. Making large-scale SVM Learning Practical. Advances in Kernel Methods-Support Vector All Rights Reserved 479