Image processing method for panoramic vehicle safety system based on multi-camera self calibration

Abstract

The invention discloses an image processing method for a panoramic vehicle safety system based on multi-camera self calibration. The image processing method relates to automatic recognition, automatic correction, automatic view angle conversion and automatic splicing image conversion. The image processing method specially comprises the steps that calibration cloth recognition is conducted on collected images, and the position information of edge corner points on checkerboard calibration cloths of the collected images are extracted automatically; a distortion model is established, distortion parameters are iterated, and a distortion correction map is obtained; affine transformation is conducted on the distortion correction map to obtain images after view angle conversion; the images collected by a plurality of cameras are subjected to transformation processing and marked by encoded points, and the images collected by the cameras are spliced and integrated. Compared with the prior art, the image processing method for the panoramic vehicle safety system based on multi-camera self calibration has the following advantages that the method supports splicing of any multiple cameras, splicing is convenient and fast, a user only needs to input the number of the cameras, the calibration cloths are matched for use, and a panoramic image can be spliced in a full-automatic mode.

Description

Technical field [0001] The invention relates to the technical field of automotive electronics, and in particular to an image processing method of a panoramic vehicle safety system based on self-calibration of multiple cameras. Background technique [0002] With the development and technological advancement of the electronic information industry, equipment capable of acquiring and recording video information has become increasingly popular. However, compared with the field of view of the human eye, the field of view of ordinary cameras is much smaller. How to effectively use computers to expand the camera The field of view of the captured images and videos has attracted wide attention from researchers. Image stitching technology can solve the limitation of viewing angle and size of imaging instruments such as cameras. The so-called image stitching is to construct a seamless, high-definition wide-view image or 360-degree panoramic view technology by aligning a series of spatially ...

AI Technical Summary

Problems solved by technology

In addition, in its affine transformation, it is necessary to substitute information such as pitch angle and height when the camera is installed, and the process is cumbersome.

[0004] Patent application document CN201310276688 only proposes a new splicing method but does not make a user experience system, and the black and white grid splicing method it uses is different from this patent

And the seam adopts the method of weight stitching. This method cannot achieve seamless stitching when the image stitching itself is inaccurate. It only weakens the gap and cannot achieve no blind spots.

Images