A method for creating distance maps based on monocular machine vision

Abstract

The invention discloses a method for creating a distance map based on monocular machine vision, which includes: extracting the red component of an image collected by a vehicle-mounted camera to obtain a reference image; Hough transform is used to detect the boundary line at the discontinuous point in the boundary; the Hough transform is performed on the result of the above boundary detection to find the straight line existing in the image, and then the endpoints of the multiple straight lines close to the vanishing point are statistically weighted, and a group of detected The average value of the column coordinates of the straight line is used as the column coordinates of the disappearing point of the camera optical axis, and the average value of the row coordinates is used as the row coordinate position of the vanishing line; the distance information is filled from the optical axis projection line to the boundaries on both sides respectively, and the scene farther than the vanishing line is calculated. Approximate processing eliminates the influence of long-distance errors on defogging processing, and obtains the scene distance and depth map. The invention solves the problems of difficulty in creating a distance map and expensive distance measuring equipment when the model is defogged.

Description

technical field [0001] The invention relates to the technical field of image processing of vehicle-mounted assisted driving systems, in particular to a method for creating a distance map based on monocular machine vision. Background technique [0002] At present, the research on driver assistance systems is mostly focused on the automatic driving of vehicles in fine weather, and there is little research on scene restoration in bad weather. The research on image dehazing at home and abroad mainly focuses on the following methods: 1. Histogram equalization algorithm; 2. Dehaze image based on color constancy theory to restore color and contrast; 3. Restoration method based on atmospheric scattering model. [0003] Among them, the histogram equalization algorithm is a commonly used simple and effective enhancement method. It is based on probability theory and corrects the histogram of the image through grayscale mapping to make it have a flat distribution. The haze degraded ima...

AI Technical Summary

Problems solved by technology

[0008] Aiming at the special structure of the car road, the present invention proposes a distance map creation method based on monocular machine vision. Through the calibration of the vehicle-mounted camera, the corresponding relationship between the scene and the CCD target surface is established, and the difficulty in creating the distance map when the model is defogged is solved. , the problem of expensive distance measurement equipment

Images