Stereo vision system enhancement method and stereo vision system

Abstract

The invention provides a method for enhancing a stereo vision system and a stereo vision system. Firstly, a plurality of original images taken by a stereo vision system for the same visual scene are obtained and original parallax images are obtained; The difference corresponds to the pixel to obtain the background pixel and establish the feature model of the background. By extracting the parallax value corresponding to the pixel segment of the continuous invalid parallax value, the hole pixel is obtained and the feature model of the hole is established. By extracting the neighborhood parallax value, the continuous effective The parallax value corresponds to the pixel segment to obtain the effective pixels in the neighborhood and establish the feature model of the neighborhood. By measuring the similarity of the three feature models and selecting the parallax value to fill the hole pixel, the present invention can accurately distinguish the hole pixel as Background or foreground to accurately fill the hole pixels.

Description

technical field [0001] The invention relates to the technical field of visual image processing, in particular to a method for enhancing a stereoscopic vision system and the stereoscopic vision system. Background technique [0002] Binocular stereo vision (Binocular Stereo Vision) is an important form of machine vision. It is based on the principle of parallax and uses imaging equipment to obtain two images of the visual scene from different positions. By calculating the position deviation between corresponding points of the image, it can A method for obtaining three-dimensional geometric information of an object. Binocular stereo vision combines the original images obtained by two "eyes" and observes the difference between them, which can obtain a clear sense of depth, establish the correspondence between features, and correspond to the image points of the same space physical point in different images , through which a disparity image can be created. [0003] According to ...

AI Technical Summary

Problems solved by technology

[0003] According to different light sources, binocular stereo vision can be respectively active stereo vision (actively providing light source) and passive stereo vision (with the help of ambient light). Among them, the range of passive stereo vision is particularly wide, which can be between 0.5m and 20m. After two original images, in the depth calculation based on stereo matching, due to the different shooting angles of the two original images, the visible background pixels in one original image may be blocked by foreground objects in the other original image And there is no corresponding point that can be matched, resulting in a "hole (gap)" in the parallax image. In the prior art, the neighborhood effective parallax value method is used to fill the "hole", but in most cases, the "gap" " is the background in the visual scene, and the disparity value of the background may not be found in the neighborhood of the "hole", resulting in the inability to correctly fill the "hole" in the disparity image

Images