Support Vector Classification For Sketch Generation Of Eyes And Brows

Sketch generation is one of the most significant fields in sketch understanding. The main aim of sketch understanding is to generate the correct semantic explanation of sketches. Sketch generation, a reverse process of sketch understanding, is based on semantic explanations and represents them in the form of hand-drawing sketch.Support Vector Machine (SVM) is a currently popular machine learning algorithm. Based on Statistical Learning Theory (SLT), it can efficiently solve small-samples machine learning problems, and has been successfully applied in many fields such as human face recognition, sketch understanding, image searching and so on.In this thesis, we introduce advanced technology of sketch generation and methods of pattern classification by Support Vector Machine, construct a Support Vector Classification model that is used to generate the sketch of eyes-and-blows, and enhance classification accuracy through parameter selection.Followings are the main works:1. Distill eigenvector from images of eyes-and-blows and construct models for sketch generation..2. Make the pretreatment of eyes-and-brows images before training. Implement a classification system which is used to correctly classify the styles of eyes-and-blows by the software pack LibSVM, and make the results of classification as the semantic explanation of sketch generation.3. Select best parameters through combining simple validation and k-fold cross-validation. Improve the accuracy of classification and demonstrate results of experiments.