A new approach of adaptive edge detection based on GAP predictor

Edge is one of the most fundamental features. Edge detection is an important method to extract the image´s characteristics. Firstly gradient adjusted predictor was used to predict the image, and then aimed at the gained error image, a new adapted approach to selected threshold based on gradient mean-value histogram was proposed, and classified edge and non-edge by the gained threshold; to obtain single pixel edges, the gained edge image was far thinned by the thinning algorithm, and the last edge image was achieved. Simulation results show that have fine and continued edge characteristic, higher precision of localization by using the proposed method compared with other methods. The experimental results are more satisfactory.