A Projection-Based Segmentation Method of License Plate Characters

Abstract

A projection-based license plate character cutting method belongs to the technical fields of computer vision and intelligent transportation. It first cuts the license plate horizontally and then cuts vertically. When re-segmenting the characters with adhesion, the selected character segmentation width is the average width of the character images that have been cut and determined to have no adhesion. After processing, if there are more than 7 character images, extract the 7 characters closest to the middle position in the license plate image, because the two sides may be interfered by stains and mistaken for characters, which greatly improves the accuracy of character segmentation; Processing, repositioning according to the average width of the other 6 characters to obtain the first Chinese character image, which can avoid the situation where the left and right radicals of the Chinese character cause the middle of the character to break.

Description

technical field [0001] The invention belongs to the technical fields of computer vision and intelligent transportation, and in particular relates to a method for cutting license plate characters based on projection. Background technique [0002] With the continuous development of the field of intelligent transportation, it is currently necessary to accurately identify license plates in road traffic monitoring, automatic toll collection systems, community vehicle management systems, and safety monitoring. License plate recognition technology generally recognizes each character separately, so it is necessary to accurately cut the license plate characters before recognition. [0003] Currently there are many methods for license plate character cutting, wherein the technical scheme closer to the present invention is: patent (Xu Yijie. Application number: 201310263494.8, title: a grayscale histogram based Figure II valued license plate character segmentation method) proposed a m...

AI Technical Summary

Problems solved by technology

[0004] To sum up, the current method of license plate character segmentation has the following deficiencies: (1) Due to the influence of stains on the license plate, the character segmentation error is caused, and the anti-interference ability is weak; (2) The situation of the license plate character adhesion is not considered, and it is easy to be segmented incorrectly ; (3) For Chinese characters with left and right radicals, it is easy to be divided into two or more characters; (4) The calculation is complex and time-consuming; (5) The position derived from prior knowledge is not accurate, resulting in inaccurate character segmentation

Images