This report investigates the development of a generalized boundary detector for two-dimensional images. Three types of boundary detection (intensity, color, and texture) together give a complete segmentation which does not depend on the type of input image. The Malik-Perona texture segmentation algorithm, which takes a multi-scale, multi-channel filtering approach, forms the initial basis of the texture implementation for this report. Good segmentation results are obtained using this approach. Several variations on the algorithm and their effects on the segmentation are presented, wit good results for many variations. The color segmentation algorithm follows essentially the same format as the last several steps of the texture algorithm, with the three color bands of an RGB image serving as the individual channels. The results for this method are quite good and often work on images for which intensity differences by themselves are not useful for segmentation. Variations are applied to the basic color algorithm to show that they maintain the effectiveness of the boundary detection. The third type of segmentation finds boundaries based upon intensity differences; the basic Canny Edge Detector is used for this purpose on several images. In fact, the Canny Edge Detector is a common denominator among the three boundary detection components, as it serves in a modified form as a back-end edge operator to the texture and color algorithms. Each of the segmentation processes reduces to the detection of edges in an intensity image after preprocessing stages.