An Approach: Modality Reduction and Face

-NN classifier for better to much harder than normal face recognition based on photo images. To recognize a face sketch through photo database, it is necessary to reduce the modality difference between face photo and face sketch images. Every face recognition system need to perform mainly three subtasks: face detection, feature extraction and classification. But face sketch recognition system through face photo database need to perform mainly four subtasks: face detection, modality reduction, feature extraction and classification. In this paper, we have concentrated on three subtasks: modality reduction, feature extraction and classification. For modality reduction, it is necessary to bring all the face images in training set and all the face sketch images intesting set toward the new dimension. We have used efficiently computable 2D discrete Haar transform at scale 3 and a negativeapproach for modality reduction. We add an integer (I) to everyimage in testing set to get better result of modality reduction on new dimension. After that, we directly applied Principle Component Analysis on newly dimension images for dimensionality reduction or feature extraction. At last, SVM and K-NN classifiers have been used for classification or recognition task. The experiment has been conducted with 100 male and female face images as training set and 100 male and female face sketch images as testing set collected from CUHK training