Abstract: Image edge detection is studied for identification in digital image brightness changes obviously, and improving the visual recognition ability of graphics. The pixel color information transform harmonic image uneven, edge detection is difficult. Contrarian and the traditional arithmetic of image edge detection using tensor model algorithm, due to the structure and operation of the tensor vector is relatively simple, edge detection performance is not good. Proposed a color tensor form of the tune and the image edge detection algorithm based on color tensor model, firstly construct the harmonic images, with the help of tensor rich operations to more accurately measure the relationship between pixels, in tensor form space for image derivative of the maximum and minimum value to detect the boundary, the improved algorithm for color tensor morphology operators based on full order reference. The simulation results show that, using this algorithm to tune and the image edge detection, it has a higher harmonic average, using tensor model to depict the color information can not only take into account the correlation between color components, but also consider the color information transformation characteristics of superior performance in detection.