Angielski
Computer vision seeks a process that starts with a noisy,§ambiguous signal from a TV camera and ends with a high-level§description of discrete objects located in 3-dimensional§space and identified in a human classification.§This book addresses the process at several levels. First to§be treated are the low-level image-processing issues of§noise removaland smoothing while preserving important lines§and singularities in an image. At a slightly higher level, a§robust contour tracing algorithm is described that produces§a cartoon of the important lines in the image. Thirdis the§high-level task of reconstructing the geometry of objects in§the scene.§The book has two aims: to give the computer vision community§a new approach to early visual processing, in the form of§image segmentation that incorporates occlusion at a low§level, and to introduce real computer algorithms that do a§better job than what most vision programmers use currently.§The algorithms are:§- a nonlinear filter that reduces noise and enhances edges,§- an edge detector that also finds corners and produces§smoothed contours rather than bitmaps,§- an algorithm for filling gaps in contours.
