A novel scheme for image segmentation is presented. An image segmentation criterion is proposed that gathers similar pixels together to form regions and creates boundaries between two dissimilar regions. This criterion is formulated as a cost function. This cost function is minimized by using gradient-descent methods, which leads to a curve evolution equation that segments the image. The proposed method generalizes previous methods to more complex similarity and distance measures and can be applied to vector valued images such as texture and color images.