Short line filtering method for image recognition

Abstract

The invention discloses a short line filtering method for image recognition. The method comprises the steps: calculating the minimum angle change degree of line segments with different spans in an image, selecting the span of the line segment with the small span and the small minimum angle change degree as a candidate threshold value of short line filtering, selecting an area in the image, identifying all lines in the area, sorting the lines according to the span size, calculating the clutter degree of the lines in each span interval, selecting the span of the line enabling the clutter degreereduction trend of the lines in the span interval to be obvious from the candidate thresholds to serve as a short line filtering threshold, and removing the lines with the span smaller than the span threshold in the image from the image. According to the method, a reasonable short line filtering threshold value is selected according to the change rule of the line clutter degree in the image, clutter meaningless short lines in the image are filtered out, the efficiency of subsequent further image processing of the image is improved, and the influence of irrelevant objects in the image on imageprocessing is reduced.

Description

technical field [0001] The invention belongs to the field of computer vision, and in particular relates to a short line filtering method for image recognition. Background technique [0002] At present, in the field of computer vision, line detection algorithms are often used in some tasks. The straight line data is used as the underlying input of the subsequent computer vision task, which determines the reliability of the final result of the task to a certain extent. For the vast majority of line detection algorithms, a large number of short lines will be generated in the line detection due to their own algorithm or image factors, and most of these short lines will not bring benefits to subsequent visual tasks, or even It will bring interference and cause strong noise, which will affect the final task result. The existing filtering methods for short straight lines are to directly remove some of the short straight lines after the straight line detection, which is relatively...

AI Technical Summary

Problems solved by technology

For the vast majority of line detection algorithms, a large number of short lines will be generated in the line detection due to their own algorithm or image factors, and most of these short lines will not bring benefits to subsequent visual tasks, or even It will cause interference and cause strong noise, which will affect the final mission result

The existing filtering methods for short straight lines are to directly remove some of the short straight lines after the straight line detection. The method is relatively direct and rough, and lacks systematic and scientific analysis.

For different line detection algorithms and images with different resolutions, the distribution of short lines will be different. Directly filtering short lines in this way may not meet the task requirements and affect the accuracy of the final result of the task.

Images