We present an image segmentation algorithm based on inscribed circle (ICseg) in this paper. The algorithm includes multi-step. It examines edges at first, and then produces a binary image with edges as foreground. After that, inscribed circles are created to cover the background of the binary image. In the first partition, the image is subdivided into separated regions by combining inscribed circles. Finally, the regions are merged by computing their shape and gray level features. This approach integrates the boundarybased and region-based techniques. It is a simple and efficient way for full image segmentation.