Pyramidal Edge Detection And Image Representation

摘要: We present a scheme of extracting edge information from parallel spatial frequency bands. From these we create an integrated image most significant edges different scales. The bands are realized using the formalism Gaussian pyramid in which levels represent bank lowpass filters. is created top-down algorithm, starting smallest version image. sequential algorithm uses mutual two consecutive to control processing lower one. This detection constitutes image-dependent nonuniform scheme. Computational results show that only 20%-50% operations needed pyramid, compared number required regular proposed generic can be also implemented with operators other than detectors exploit advantages inherent biological images.