Abstract
This paper presents an efficient color filter (CF) defect recognition system, based on gray-level co-occurrence matrix (GLCM) and support vector machines (SVM). GLCM is used to extract textural information first. The extracted features are further selected by F-score, information gain and individual feature accuracy selection methods. The idea is to reduce dimensionality of CF defect space. The selected features are used as features for the SVM classification process. Experimental results indicate the proposed method achieves outstanding performance for CF defect classification.