A stereo image matching method based on a convolutional neural network

Abstract

The invention relates to a three-dimensional image matching method based on a convolutional neural network, which comprises the following steps: acquiring a plurality of three-dimensional matching image pairs and corresponding real disparity, and taking the three-dimensional matching image pairs and the corresponding real disparity as a data set; Constructing a convolutional neural network, and selecting a linear correction unit (RELU) function for activation; Training the convolutional neural network by adopting a back propagation algorithm, and determining a network error function and a learning rate; Through the calculation of the convolutional neural network, the network outputs a matching cost space diagram of the left and right image blocks; And performing matching cost aggregation,disparity selection and disparity refinement on the cost space graph, and selecting a pixel point with the minimum cost as a matching point to obtain a final disparity map. The image can be directly used as network input, the overall disparity map obtained through the convolutional neural network matching algorithm is relatively smooth, a relatively good matching effect can be achieved in a non-texture region and a depth value abrupt change region, and relatively good robustness is still achieved for an image pair with illumination change and incomplete correction.

Description

technical field [0001] The present invention relates to the technical field of three-dimensional video, in particular to a stereoscopic image matching method based on a convolutional neural network. Background technique [0002] Stereo matching techniques are now widely used in problems such as robot navigation, depth-of-field rendering, and video processing. The binocular stereo matching in the stereo matching technology generally acquires two images of the same scene from different angles, and calculates the parallax information, and then obtains the three-dimensional depth information of the object according to the parallax through the principle of triangulation. [0003] For the human eye, it is very easy to restore the three-dimensional information from the scene, but for the computer, it is extremely difficult to find the matching corresponding point of the same scene point in the binocular image pair, and the stereo vision technology is still very difficult. It is no...

AI Technical Summary

Problems solved by technology

The main consideration in stereo vision is the semi-occlusion phenomenon. Since there is no corresponding matching point for the occlusion point, the parallax of the occlusion point can only be estimated by the parallax of the surrounding points.

Images