A novel algorithm for the image transition region extraction and segmentation is presented based on local definition-complexity of the image.At first,the local definition of the original image is calculated to get a definition image,which enhanced the transition region′s grey level.Then the local complexity of the definition image is calculated and transition region extraction threshold is gotten based on the complexity curve.At last,segmentation is made to the original image according to the threshold value obtained from the extracted transition region histogram.Experiment results demonstrate that the proposed algorithm achieves a better transition region extraction and segmentation performance.