Edge detection using advanced grey prediction model

Conventional image edge detection algorithms generate the information undetected and artificial information problems. In order to detect the image edge more effectively, an improved method based on grey model (GM) is brought forward. The neighborhood pixels of target pixel are selected to build the model. These data are preprocessed by translation transformation and logarithmic transformation. The prediction image is gained by the improved GM(1,1) with enhancement parameter p. The original image subtracted from prediction image gives the edge of image. With the improved algorithm, 80 standard images with bmp format 256*256 resolution are processed, and the result shows that not only the new algorithm gain more precise and useful edge pixel, but also the details of the image are obtained. The improved algorithm has better performance than the original algorithm.