Self-Adaptability Research of Billet Character Segmentation in the Complex Scene

Abstract

In the process of billet detection and recognition, the self-adaptability research of billet character segmentation in the complex scene is a highly complicated intelligence problem. On account of a good many of factors, such as the uneven ends of billet, the rusty billets, the non-standard printed characters and so on, the irregular characters at ends of steel billet may be connected and broken. Improper segmentation of connected and broken characters will lead the characters distorted directly and then impact on the recognition results. In view of these complex factors, a segmentation algorithm based on self-adaptability decision is proposed to improve the segmentation self-adaptability in this paper. The segmentation experimental results show that the proposed algorithm divides the billet characters in the complex scene accurately and improves the self-adaptability of character segmentation.