A fast disparity image calculation method and device

Abstract

The invention provides a fast anaglyph calculating method and device. The fast anaglyph calculating method and device are applied to image processing equipment. The method comprises the steps that background subtraction processing is carried out on a pair of original images which are output by a binocular camera and correspond to each other in time, and a pair of background images and a pair of foreground images are obtained; a background anaglyph is generated according to a preset first parallax algorithm based on the background images; a foreground anaglyph is generated according to a preset second parallax algorithm based on the foreground images; the foreground anaglyph and the current background anaglyph are integrated into a complete anaglyph. Compared with the prior art, the real-time performance of complete anaglyph calculation can be greatly improved, and prominent technical significance in monitoring and other professional application markets is achieved.

Description

technical field [0001] The present application relates to image processing technologies, and in particular to a fast parallax image calculation method and device. Background technique [0002] With the in-depth development of multimedia, Internet and other technologies, ordinary two-dimensional mode images / videos are increasingly unable to meet people's needs. In recent years, three-dimensional (3D) images have been used more frequently. When people watch 3D movies, they can obtain a stereoscopic visual experience that is difficult to achieve with 2D movies. In addition to the application in film and television, 3D information close to the real scene has gradually been used in various other fields, especially in the field of video surveillance, where the stereo vision system can more effectively locate, distinguish or track the monitored target. In common monitoring services, such as detection or tracking of vehicles and pedestrians, three-dimensional video information can...

AI Technical Summary

Problems solved by technology

The feature-based method has better anti-interference performance, but it can only obtain a sparse disparity field, and to obtain a dense disparity field must be supplemented by a complex interpolation process

The area method can generate a dense parallax field. The algorithm is simple, effective, and easy to implement. The disadvantage is that the accuracy is not high and it is easily affected by noise.

However, the calculation speed of this scheme is faster, but it needs to rely on an initial parallax estimation. If the initial estimation error is obvious, the effect after dynamic programming is not ideal.

[0005] 2) Global matching algorithm, this type of method utilizes the global constraint information of the image, defines an energy function including data error, smoothness and occlusion on the entire image, and converts the problem into an optimization problem to solve. The image is blur insensitive, but it is computationally expensive

Slow processing speed may not be acceptable to users in professional fields such as monitoring

Images