Binocular stereoscopic vision parallax filtering method and device based on self-supervised learning

Abstract

The invention relates to a binocular stereoscopic vision parallax filtering method and device based on self-supervised learning, and the method comprises the steps: 1, taking a pixel point P from a to-be-processed image, taking a rectangular region with the pixel point P as the center, adding the offset d to the coordinates of the pixel point P, obtaining a pixel point M, taking a rectangular area with the same size on the same image by taking the pixel point M as the center; 2, repeating the step 1 to obtain two rectangular regions as a pair of training samples, and taking d1 as a label of the sample pair; 3, repeating the step 1 and the step 2 to obtain a certain number of training sample pairs and corresponding labels; 4, inputting the training sample pair and the corresponding label in the step 3 into a convolutional neural network model, and calculating a loss function to obtain the convolutional neural network model; and 5, judging whether the two sample pairs are matched or not, and if the two sample pairs are not matched, determining the pixels as noise points and filtering the noise points.

Description

technical field [0001] The present application relates to the field of computer vision, in particular to a binocular stereo vision parallax filtering method and device based on self-supervised learning. Background technique [0002] Computer vision is a technology about how to use cameras and computers to obtain the data and information needed for production and operation. With the development of computer vision technology, computer vision has been widely used in various scenarios. Binocular stereo vision simulates the human visual system and is an important branch of computer vision. Binocular stereo vision technology is based on the principle of parallax and uses imaging equipment to obtain two images of the measured object from different positions, and obtains the three-dimensional geometric information of the object by calculating the position deviation between the corresponding points of the image. At present, the matching process in binocular stereo vision is prone t...

AI Technical Summary

Problems solved by technology

[0004] The embodiment of the present application provides a binocular stereo vision parallax filtering method, device, electronic device and storage medium based on self-supervised learning, to at least solve the problem of three-dimensional Problems with noise in the point cloud

Images