Scratch detection method suitable for vehicle distant view image

Abstract

The invention provides a scratch detection method suitable for a vehicle distant view image. The scratch detection method comprises the following steps: segmenting a region of interest where scratchesmay occur by using a deep learning algorithm; obtaining a candidate scratch region in the region of interest by using scratch detection fusing color and spatial information and an MSER method; and screening the candidate scratch areas by comprehensively utilizing Hough line detection and an SVM method, and marking the scratch areas. According to the method, multiple image processing methods are comprehensively used for detecting the scratch area in the interested area, and the problems that features are not obvious and scratches and interference areas are difficult to distinguish due to the fact that the area of the scratch area in the vehicle distant view image is too small are solved. Meanwhile, close-range images do not need to be shot manually, and manual operation is liberated from complex scratch detection tasks.

Description

technical field [0001] The present invention relates to the fields of image processing and deep learning, in particular, to a scratch detection method suitable for vehicle perspective images. Background technique [0002] Currently, there is a lack of research dedicated to the detection of scratches on vehicle surfaces, and a similar one is research on vehicle damage detection. Existing vehicle damage detection methods can be mainly divided into two categories: computer vision-based damage detection methods and deep learning-based damage detection methods, and vehicle damage detection methods based on deep learning can be further divided into detection methods based on residual dense networks And a detection method based on the Faster R-CNN target recognition algorithm. The main ideas of various methods are as follows: [0003] (1) Vehicle damage discrimination method based on computer vision: This method first needs to arrange a binocular image acquisition system, and use...

AI Technical Summary

Problems solved by technology

[0008] According to the above-mentioned technical problems of high cost and poor detection accuracy of the existing technology, a scratch detection method suitable for vehicle vision images is provided

The invention comprehensively uses a variety of image processing methods to detect the scratch area in the region of interest, avoiding the problem that the scratch area in the vehicle vision image is too small to have obvious features and difficult to distinguish the scratch area from the interference area

Images