A Novel SVMs Classifier Based on Fourier Descriptor and Other Multi-features Fusion

According to the global and local features of images, Fourier descriptor and other multi-features is introduced for SVMs classifier. At first, extracting features of images is done, then classification method of SVMs for recognition is discussed. Experimentation with 11 image groups is conducted and the results prove that Fourier descriptors are simple, efficient, and effective for recognition of images, and the SVMs method has excellent classification and generalization ability in solving learning problem with small training set of sample. The comparison of different kernel functions for SVMs shows that linear kernel function is most suitable for image recognition, and the best recognition rate of 98.5% of one image group is achieved.