Performance Evaluation of SVM in Image Segmentation

Traditional classification methods, such as neural network approaches, have suffered difficulties with generalization and producing models. Support vector machine (SVM) approach is considered a good candidate because of its high generalization performance without the need to add a priori knowledge, even when the dimension of the input space is very high. In this paper, SVM approach is proposed to segment images and we evaluate thoroughly its segmentation performance. Experimental results show that: (1) the effect of kernel function, model parameters and input vectors on the segmentation performance is significant; (2) SVM approach is suitably used as learning machine under the condition of small sample sizes; (3) SVM approach is less sensitive to noise in image segmentation.