Wavelet-based multiresolution edge detection utilizing gray level edge maps

摘要: Edge detection is an important low-level vision problem. Most edge methods operate on image at a single resolution and output binary map. Edges within image, however, generally occur various resolutions, or scales, represent transitions of different degrees, gradient levels. Thus, that maps do not always yield satisfactory results. This paper develops multiresolution method utilizes multirate wavelet decomposition to generate series images with progressively lower resolution. are then recursively extracted form where the restricted be set reflect level each point. The stacking map pyramid. In this formulation, base (lowest level) contains edges all scales while pruning, based scale, performed subsequent approach shown have advantages over previously defined methods. Results presented using natural seen images. These results motivated by tactile imaging, problem in making visual information accessible blind individuals.