A Parallax Image Fusion Method for Multi-band Stereo Camera

Abstract

The invention discloses a parallax image fusion method of a multi-band stereo camera, which belongs to the fields of image processing and computer vision. The method proposed by the present invention uses the intermediate output of binocular disparity estimation to obtain pixel disparity confidence information. The confidence information can be used to judge the position disparity reliability and assist disparity fusion. The confidence degree acquisition process makes full use of the intermediate output of the calculation, which can be easily embedded into the traditional disparity estimation process, with high calculation efficiency, simple and easy operation. In the multi-band stereo camera parallax fusion method proposed by the present invention, the disparity map involved in fusion is obtained from binocular images of corresponding bands, fully utilizes the information of each band, and avoids introducing uncertainty and errors at the same time. According to the strategy of disparity fusion based on confidence, the accurate information of each band can be selected to achieve the purpose of complementary advantages of multi-band equipment.

Description

technical field [0001] The invention belongs to the field of image processing and computer vision, relates to multi-modal binocular parallax estimation and confidence degree calculation, and performs fusion and depth calculation of parallax information obtained by multi-band sensors. In particular, the invention relates to a confidence calculation method and a parallax image fusion method of a multi-band stereo camera. Background technique [0002] The development of binocular stereo vision technology based on visible light band is relatively mature. Visible light imaging has rich contrast, color, and shape information, so it can accurately and quickly obtain matching information between binocular images, and then obtain scene depth information. However, imaging in the visible light band has its defects. For example, in strong light, fog, rain, snow or night, the imaging quality is greatly reduced, which affects the matching accuracy. Therefore, using the complementarity of...

AI Technical Summary

Problems solved by technology

The above method matches binocular cameras with different bands separately, and then maps the images into three-dimensional information according to the matching results, and performs fusion on the point cloud level, which greatly increases the amount of calculation and resource consumption, and is difficult to apply to actual products

Images