Method for obstacle segmentation and device thereof

Abstract

The invention relates to a method for obstacle segmentation and a device thereof. The method for obstacle segmentation comprises the following steps: current frame images in sequential images and the adjacent frame images with frame interval being larger than or equal to 1 frame are divided into a plurality of identical blocks, the difference is carried out on the image blocks at the corresponding positions of the current frame images and the adjacent frame image, and differential block areas in the differential images are determined; the verticality and the texture of all the corresponding blocks in the current frame images and the differential block areas are calculated; the divided blocks which are corresponding to the differential block areas in the current frame images and have verticality and texture are determined into obstacle blocks; and the obstacle blocks generate obstacle areas according to the communicating regulation. By using the invention, the aim of accurately segmenting the obstacles in the image can be realized under the dislocation state of the camera.

Description

technical field [0001] The present invention relates to the technical field of image computing, in particular to an obstacle segmentation method and device. Background technique [0002] The difference image method, that is, the frame differencing algorithm, is an image calculation method for judging whether there is a moving object in the sequence image and detecting the moving object based on the difference analysis between the current image and the reference image (background image). [0003] Based on this obstacle segmentation algorithm, through image segmentation, find image areas that may contain obstacles in the image, and then identify whether these areas are obstacles. In this field, an obstacle generally refers to a three-dimensional object above the ground, which has vertical features and features that are different from the background. [0004] At present, the image difference method is a main obstacle segmentation algorithm based on monocular vision, and its pr...

AI Technical Summary

Problems solved by technology

Moreover, this method requires background modeling and background updating process, and can only segment moving obstacles

[0005] However, when the camera moves, the background image will change. In this case, the method will be used to segment a large number of non-obstacles into obstacles, such as segmenting a large number of non-obstacles in the background of the road that moves in the camera into Obstacles, such as road surface cracks, paint, etc., lead to increased subsequent obstacle identification processing time, as well as the possibility of false identification, so the practicability is not high

Images