Study on Extracting Edge of Cropland Scenery

Vision navigation and location is a main function in the vision system of intelligent agricultural mobile robot, and the edge feature of images is an important feature for vision navigation and location. According to the characteristic of cropland scenery, a compactly supported dyadic antisymmetric wavelet with respect to origin is brought forward to detect edges of cropland image. A set of filters were given that can be used to construct the edge detecting wavelet. The edge features are provided by determining the local maxim of wavelet coefficient at dyadic scale of the image. After the computer simulation was carried out on the cropland image, the continuous and smooth edge image can be got. The edges of cropland scenery are extracted accurately. The experiment result reveals that the method is efficient and practicable.