Image processing method and device and terminal

Abstract

The invention discloses an image processing method and device and a terminal, and belongs to the technical field of automobile assistant driving. The method comprises that a V disparity map corresponding to a parallax image to be detected is obtained, pixel points in each column of the V disparity map are arranged, and a type attribute of each column is determined according to a statistical result; and on the basis of the type attribute of each column, pavement candidate points are selected from corresponding pixel points according to a preset rule, and points which cause interference in detection of an obstacle are determined and deleted on the basis of the selected pavement candidate points. The type attribute of each column can indicate whether pixel points of the corresponding column include points corresponding to a pavement mark line or the obstacle, namely, the type attribute of each column can indicate distribution rules of parallax points corresponding to the pavement and obstacle in the V disparity map in practical; and thus, the pavement candidate points are selected from each column of pixel points according to the type attribute, and the pavement candidate points obtained by selection are more fit to the pavement distribution rules.

Description

technical field [0001] The present invention relates to the technical field of automobile driving assistance, in particular to an image processing method, device and terminal. Background technique [0002] With the continuous development of intelligence, the auxiliary driving technology of automobiles has become a technology hotspot that automobile manufacturers are chasing. In existing advanced driving assistance systems, binocular cameras can be used to collect road condition images, and the collected road condition images can be processed through a stereo matching algorithm to obtain parallax images corresponding to the road condition images. Afterwards, obstacles ahead can be detected in this parallax image to ensure safe driving of the car. Wherein, in order to ensure the accuracy of obstacle detection, before determining the obstacle ahead in the parallax image, the parallax image can be processed first, so as to eliminate the points corresponding to the road marking ...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problem in the related art that the points corresponding to the road surface determined when processing the parallax image tend to have large deviations, thus causing the problem that the parallax point in the parallax image is accidentally deleted or omitted, an embodiment of the present invention provides an image Processing method, device and terminal

Images