Dissertation > Excellent graduate degree dissertation topics show

Research on Application of Fractional Differential to Digital Image Processing and Support Vector Machine to Face Recognition

Author: LanLiBin
Tutor: WangChengLiang
School: Chongqing University
Course: Applied Computer Technology
Keywords: Fractional differential Image texture enhancement Image Interpolation Face Recognition Support Vector Machine
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 192
Quote: 1
Read: Download Dissertation

Abstract

The digital image processing method derived from two main areas of application: one is the improvement of the image information in order to facilitate the people to analyze; second is to make the machine automatically understand the identification of the image data storage, transmission and display. Fractional differential for image processing, both at home and abroad is a little research in the emerging field. Compared with the traditional image processing method based on integer-order differential fractional differential for image processing has obvious advantages. Face recognition is an important application of pattern recognition research in recent years become a hot research questions support vector machine for face recognition. The fractional differential for image processing and support vector machine for face recognition, the main work and results are as follows: First, a more systematic analysis and summarizes the basic theory of digital image processing, including digital image processing the concept of the research status and significance of the traditional digital image processing technology, and research status of fractional differential image processing; study and discusses the theory of fractional calculus, including the course of development of the theory of fractional calculus, three typical fractional calculus defined and fractional calculus in the basic theory of special functions. At the same time, a more systematic analysis and summary of the basic theory of face recognition, including a variety of face recognition technology, face recognition technology research status and support vector machine in face recognition status; study and discusses the The basic theory of support vector machine, including support vector machine research progress, the background theory and its basic theory, basic theoretical knowledge. Second, proposed a theory based on the size of the mask window, GL formula, image gradient features and human eye visual characteristics can automatically generate fractional differential order adaptive mathematical functions, and according to the design of the operator mask , in without human specify an optimal fractional differential order in the case, the theory of fractional differential can be fully automated image processing, saves a lot of time to find the best artificial fractional differential order some extent to meet a large number of dynamic The sequence image enhancement processing requirements. Information entropy, average gradient image texture features evaluation parameters for quantitative analysis and experimental verification. Experimental results show that the method can be obtained for any gray-scale image enhancement effect of continuous change, close to the best fractional differential enhancement, in line with the perception of the people, is an effective method of image texture enhancement. The edge interpolation algorithm for the current interpolated image can not effectively improve the low-frequency texture detail, study and propose an image interpolation algorithm based on fractional differential edge detection. Designed and implemented based on fractional differential theory, can effectively extract the low-frequency texture information operator mask. , In accordance with the texture information of the detected edges, respectively, along the edge direction, perpendicular to the edge direction and a smoothing region to be interpolated pixels linear interpolation, quadratic interpolation and bilinear interpolation. With a peak signal-to-noise ratio (PSNR) and information entropy (IE) image quality evaluation criteria for quantitative analysis and experimental verification. The results show that the method can be rich texture information, and improve the peak signal-to-noise ratio and the results were consistent visual experience of the people. Fourth, in order to improve the recognition accuracy of face recognition system to study the face recognition method based on PCA feature and SVM classifier, and using PCA NN, SVM recognition methods on ORL face database, the experimental comparison. Experimental data show that face recognition method based on PCA feature and SVM classifier in the case of small sample, the recognition rate is better than PCA NN SVM recognition methods, which indicates that the PCA feature of the face samples fed into the SVM classifier to classify and recognize the feasibility and correctness. Light and expression of the face image, the extracted based on fractional differential binary edge image and into the support vector machine classification method, comparative analysis based on binary edge image based on gray-scale image recognition effect and experiments were carried out in ORL and Yale face database. The experimental data show that based on fractional differential extraction of binary edge image, illumination and expression changes is robust, more conducive to the classification and identification.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net Mobile