Based On The Least Squares Estimate Facial Sketch Effect Algorithm Research

Cartoons and cartoon effects image processing is a realistic performance practices based on art painting. With the rapid development of the computer industry, the use of the cartoon industry developed by the computer is an auxiliary authoring tool and it developed the entire industries in recent years. At the same time, the human face has been one of the focus research fields in academic circles all over the world, as well as face image video, and cartoon style image processing is one of the focus research in academics.In this thesis, we select the face as object of study, aimed at generating the sketch-style cartoon face under an intelligent cartoon algorithm, this algorithm makes real photographed face image into sketch styles of cartoon images. The general process of the proposed algorithm can be divided into two steps:the least squares estimation and edge detail enhancement.The main research work includes:(A) In the first step, the main study is with the premise of the face image samples. The problem is how to linearly present an unknown real face image with sample face images, and its solution is the least squares error estimate. The real face image library get the coefficient vector which is mapped to a cartoon face database to enable a cartoon style face and find the real face images which is closest to the cartoon one. Our method can guarantee a good surface witness, no easy distortion symmetry and no contour distortion.(B) In the second step, the main research is how to enhance the edge of the chrominance information for grayscale face images. The method includes bilateral filter, the removal of noise in overall intra-regions, DoG edge detection algorithm.(C) A set of standard face images and a group of sketch images drawn by professional artists are used, and the experimental simulation environment is Matlab. We analyze the experimental parameters, the experimental efficiency, the various elements of the experiment and experimental comparisons.(D) On the basis of the proposed algorithm, we develop software by Microsoft Visual Stutio2008, and OpenCV2.1as API to complete opening image, face detection, image cropping, sketch generation functions.