DOI: 10.36347/sjet

Pages: 70-82

Face recognition is the most emerging research area of the pattern recognition since the early 1990s. This paper presents a new classifier called modified fuzzy support vector machine (MFSVM) by modification in membership function of Fuzzy Support Vector Machine using Combination of Distance Feature, Correlation. We use based on discrete wavelet transform (DWT), discrete Cosine transform (DCT) as a combined feature extraction method and modified fuzzy support vector machine (MFSVM) for face recognition. First the face image is decomposed by 2-D wavelet, then the 2-dimentional DCT is applied to the low-frequency image. Then, using the DCT coefficient, face image can be recognized using MFSVM classifier. The experiment is carried out on the ORL database the result is encouraging, which achieves high accuracy