Abstract: An approach to image texture segmentation is proposed based on wavelet and curve evolution.2D wavelet decomposes the texture image into four wavelet coefficients which indicate the texture feature and form 4D vector image.In order to get the segmentation result of the original image texture,the wavelet coefficient vector image is segmented and classified by using curve evolution based on the maximum posterior probability(MAP)model.Compared with wavelet FCM method,the experiments have proved that the proposed method can get better results such as that the segmented edge is continuous,and it is robust to noise.