Advances in Natural and Applied Sciences. Detecting Wrinkles and Sagginess of Skin from Human Face Photographs

Transcription

1 AENSI Journals Advances in Natural and Applied Sciences ISSN: EISSN: Journal home page: Detecting Wrinkles and Sagginess of Skin from Human Face Photographs 1 Arumugam P, 2 Muthukumar Subramanyam, 3 Selvakumar S and 4 L.Sindhu 1 Department of Statistics, MS University, Tirunelveli , Tamil Nadu, India 2 Professor, Dhirajlal Gandhi College of Technology, Department of computer science and engineering, Salem , Tamil Nadu. 3 Research Scholar, MS University, Tirunelveli , Tamil Nadu, India 4 AP, Dhirajlal Gandhi College of Technology, Department of computer science and engineering, Salem , Tamil Nadu. A R T I C L E I N F O Article history: Received 23 July 2015 Accepted 28 August 2015 Available online 25 September 2015 Keywords: Wrinkles, Sagginess, Face detection, Forehead wrinkle identification, Inpainting. A B S T R A C T The detection and inpainting of wrinkles and sagginess of human facial images is the main objective of this research. Advanced aging skin shows signs of overall sagging of skin in addition to wrinkles and rough texture. In this research the Morphological Linear Transformation (MLT) based algorithm detects wrinkles and sagginess present in the surrounding skin. The detection of wrinkles and sagginess allow the system to select skin features for processing the surrounding skin without much user interaction. Later, a new variational method of inpainting is adopted for improving the detected skin imperfections. This algorithm address the wrinkles and sagginess of the skin by selecting the best matching patch from the unaffected skin and also address the artifacts caused by repetition. Both detection and inpainting of wrinkles and sagginess are done with minimum user interaction. Thus, the methodology minimizes the interaction by the user s learning expertise AENSI Publisher All rights reserved. To Cite This Article: Arumugam P, Muthukumar Subramanyam, Selvakumar S and L.Sindhu, Detecting Wrinkles and Sagginess of Skin from Human Face Photographs, Adv. in Nat. Appl. Sci., 9(12): 59-64, 2015 INTRODUCTION Faces are the most important module of objects computers have to deal with. Automatic recognition and processing of facial images with wrinkles and sagginess is a challenge in image processing. Facial image processing is transforming an input facial image into another and involving no high-level semantic classification. Facial image processing includes typical applications like detection and removal of facial wrinkles and sagginess, eye glass removal, synthesizing facial expression and restoration of facial images. The skin of matured people becomes more prone to lines, wrinkles, sagginess and folds that become more pronounced with time. Skin wrinkles and sagginess are perceived as important cues in communicating information about the age of the person. Therefore estimation of the degree of facial wrinkling and sagging is used for assessing benefits to facial appearance due to various dermatological treatments. However, few image-based algorithms are present for computationally assessing facial wrinkles and sagginess. Wrinkles and sagginess are more likely to occur in old age and the skin becomes thinner and loses its elasticity. Production of less oil makes the skin drier and more susceptible to wrinkles; less fat in the skin makes it loose and saggy. The aging skin becomes thinner and more easily smashed, with the appearance of wrinkles and sagginess. The deterioration is also accompanied by a darkening of skin color. In the case of image processing, it is considered that the texture appearance is changing with image recording parameters, that are camera, elucidation and path of view, a difficulty common to any real surface.\ Fig. 1: Image inpainting Corresponding Author: Arumugam P, Department of Statistics, MS University, Tirunelveli , Tamil Nadu, India

2 60 Arumugam P et al, 2015 Image inpainting is the technique of modifying an image in an imperceptible form. The goals and applications of inpainting varies from the restoration of damaged paintings and photographs to the removal/replacement of selected objects. The modification of images in a non-detectable way for an observer who does not know the original image is as old as artistic creation itself. The objective of inpainting is to reconstitute the missing or damaged portions of the work, in order to make it more decipherable and to re-establish its unanimity.the basic idea of imperfections detection and inpainting is to smoothly propagate information from the surrounding areas. The restored images will be sharp without any color artifacts. Detecting the imperfections and inpainting them by avoiding the color artifacts helps to reduce the type for restoration by reducing the magnitude. Related Work: Detection of wrinkles and sagginess methods target on finding the imperfections in the structure and texture of an image. A survey of imperfections detection methods can be found in (Efros, A.A. and W.T. Freeman, 2001; Batool, Nazre, and Rama Chellappa, 2014; Batool, N. and R. Chellappa, 2012) and (Vaghela, Khyati, and Narendra Patel, 2013). A novel method called image quilting in (Efros, A.A. and W.T. Freeman, 2001) is used for seamless stitching of small patches of the exemplar texture. This method was used to stitch skin patches together to fill the gaps left by removal of wrinkled regions. Filling of gaps in images using texture synthesis is also called as constrained texture synthesis. A new generative model was proposed for wrinkles on aging human faces using Marked Point Processes (MPP) in (Batool, N. and R. Chellappa, 2012). Wrinkles are considered as stochastic spatial measures of sequence of line segments, and detected in an image by appropriate localization of line segments. Wrinkles are restricted by sampling MPP using the Reversible Jump Markov Chain Monte Carlo (RJMCMC) algorithm. The other references focus on the specific application of imperfections removal i.e. image inpainting. The image inpainting technique works by finding a suitable texture template in the image to fill in the gap with, then it calculates the seamless warping between the template and the gap and fills the gap via texture synthesis. The imperfection detection approach is based on Morphological Linear Transformation algorithm. In the following the approach is presented in detail. Approach: The affected area of the image i.e., the imperfections of the skin is detected by finding the texture and the binary mask of the image. The detected imperfections are inpainted using some algorithm which employs best matching patch technique. The best matching patch is selected pixel by pixel from the unaffected area of the image. This step should be repeated in an iterative manner until finding the best patch to be replaced.the MLT algorithm incorporating the morphological operations such as erosion, dilation, opening and closing is used for detecting the wrinkles and sagginess of the skin. This algorithm is applied on the textured and the masked image in order to get better result of imperfection detection. The binary mask of the texture image is compared with the input image in order to determine the area to be inpainted. Finally the inpainting is carried out with the suitable algorithm adopting the best matching patch technique. A. Finding the Texture Gradient and Finding the Binary Mask: An image texture is a set of metrics calculated in an image in order to quantify the perceived texture of an image. The spatial arrangement of colors or intensities of an image or a particular region of an image can be identified by its texture. Usually texture can be determined by approaches like edge detection, co-occurrence matrix, laws texture energy measures. Fig. 2: System Architecture

3 61 Arumugam P et al, 2015 The imperfections in the skin are determined by first finding the texture of the image. Here, the texture of the image is determined by using imfilter function. This function determines the texture along the direction of the energy. Also, the imfilter function converts the RGB image to a colorless i.e., grayscale image and highlights the imperfections in white color in the textured image. The binary mask of the image is found by setting some threshold value. The binary mask of the image is found in order to detect the skin imperfections like wrinkles, sagginess by using morphological operations. B. Detection of Skin Imperfections: After finding the texture and binary mask of the image, the imperfections of the skin such as wrinkles and sagginess are spotted clearly by MLT algorithm, applying the morphological operations. The closing operation is performed in the green plane of the image in order to determine the structure of the image. Similarly the erosion, dilation operations are used to perform the thickening, thinning of the image C. System Flow Diagram: The input image almost certainly the facial image of an aged person is processed with suitable filter and function, probably the imfilter function in order to determine the texture and binary mask of the image. The texture of the facial image of an aged person is used to detect the wrinkles and sagginess. These imperfections are restored by replacing the best matching skin from the unaffected area. The best matching patch is selected pixel by pixel. If the perfect patch is selected, then the patch found can be used to replace the affected area. If not, then the selection of best matching patch process is repeated until finding the best patch to be replaced. Fig. 3: System Flow Diagram Experiments And Results: The experiment is conducted by applying the MLT algorithm and the detected imperfections are resulted. Fig (a) shows the texture of the forehead image extracted using imfilter function. Fig (b) shows the application of opening, closing morphological operations to the textured image. Fig (c) shows the separation of green plane from fig (b). In fig (d) the image is positioned by setting the threshold and then the image is tilted. The positioned image and tilted image is applied to erosion and dilation morphological operations in fig (e). Finally, the wrinkles and sagginess of the image are detected in fig (f). Fig. (a): Texture Detection

4 62 Arumugam P et al, 2015 Fig. (b): Image Obtained by Applying Opening, Closing Operations Fig. (c): Separation of Green Plane from Fig. 5 Fig. (d): Positioning The Image By Setting Threshold Fig. (e): Image Obtained by Performing Erosion, Dilation Operations

5 63 Arumugam P et al, 2015 Fig. (f): Detection of Skin Imperfections Performance Evaluation: The performance of the research, detection of wrinkles and sagginess in the advanced aging skin of humans was qualitatively performed by conducting a survey among different users. The users were asked to evaluate the result obtained based on a quality scale ranging from low to excellent. Each user evaluates individually and finally the average of the evaluation based on the quality of the result obtained is taken as the performance measure. This measure determines whether the research done is qualitative or not. Table 1: Quality scale RANGE QUALITY SCALE 0-25% LOW 25-50% MEDIUM 50-75% GOOD % EXCELLENT Table 2: Performance evaluation USERS EVALUATION USER 1 75% USER 2 80% USER 3 92% USER 4 87% USER 5 69% USER 6 77% USER 7 89% USER 8 90% USER 9 87% USER 10 82% USER 11 94% Finally, the average quality evaluated as a performance measure seems to be 83.3% which is an excellent measure. This confirms that the research carried out is excellent in terms of its quality. Conclusion And Future Work: A. Conclusion: The presence of imperfections such as wrinkles and sagginess in advanced aging skin is detected. The algorithm used incorporates the morphological operations in order to detect the exact location of the imperfections automatically. The detection of imperfections can allow these skin features to be processed differently than the surrounding skin without much user interaction. With minimum user interaction, the algorithms are able to detect most of the imperfections. Overall, the algorithm used presents significant improved detection of imperfections in cases where skin imperfections are more visible in the surrounding skin. B. Future Work: This research can be extended by introducing a suitable inpainting algorithm so that the imperfections detected can be removed and the skin looks flawless. The inpainting algorithm should be selected in such a way that it should employ minimum user interaction and should provide better results than the prevailing algorithms. REFERENCES Efros, A.A. and W.T. Freeman, Image quilting for texture synthesis and transfer, in Proc. 28th Annu. Conf. Comput. Graph. Interact. Techn., pp: Efros, A.A. and T.K. Leung, Texture synthesis by non-parametric sampling, in Proc. Int. Conf. Comput. Vis., pp:

6 64 Arumugam P et al, 2015 Criminisi, A., P. Perez and K. Toyama, Object removal by exemplar based inpainting, in Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit., 2: II-721 II-728. Bugeau, A., M. Bertalmio, V. Caselles and G. Sapiro, A comprehensive framework for image inpainting, IEEE Trans. Image Process., 19(10): Arias, Pablo, et al "A variational framework for exemplar-based image inpainting." International journal of computer vision., 93(3): Batool, Nazre, and Rama Chellappa, "Fast detection of facial wrinkles based on Gabor features using image morphology and geometric constraints." Pattern Recognition. Bertozzi, Andrea L., Selim Esedoglu and Alan Gillette, "Inpainting of binary images using the CahnHilliard equation." IEEE Transactions on image processing, 16(1): Hareesh, Anamandra Sai, and Venkatachalam Chandrasekaran, "Exemplar-based color image inpainting: a fractional gradient function approach." Pattern Analysis and Applications., 17(2): Li, Shutao, and Ming Zhao, "Image inpainting with salient structure completion and texture propagation." Pattern Recognition Letters, 32(9): Bertalmio, M., L. Vesa, G. Sapiro and S. Osher, Simultaneous structure and texture image inpainting, IEEE Trans. Image Process., 12(8): Elad, M., J.-L. Starck, P. Querre and D.L. Donoho, Simultaneous cartoon and texture image inpainting using morphological component analysis (MCA), Appl. Comput. Harmon. Anal., 19(3): Batool, N. and R. Chellappa, Modeling and detection of wrinkles in aging human faces using marked point processes, in Proc. Eur. Conf. Comput. Vis. (ECCV) Workshops, pp: Georgiev, T., Image reconstruction invariant to relighting, in Proc. Eurographics, Dublin, Ireland, pp: Vaghela, Khyati, and Narendra Patel, Automatic text detection using morphological operations and inpainting. International journal of innovative research in science, engineering and technology IJIRSET 2(5): Wang, Jing, et al., "Robust object removal with an exemplar-based image inpainting approach." Neurocomputing, 123: Yu, Yibin, et al., "Fast Wavelet Thresholding Algorithms for Face Image Inpainting." Green Computing and Communications (GreenCom), 2013 IEEE and Internet of Things (ithings/cpscom), IEEE International Conference on and IEEE Cyber, Physical and Social Computing. IEEE. Tauber, Z., Z.-N. Li and M.S. Drew, Review and preview: Disocclusion by inpainting for image-based rendering, IEEE Trans. Syst., Man, Cybern. C, Appl. Rev., 37(4):