Method, system, device and medium for automatically identifying vehicle-mounted scrap steel in full load state

By employing deep network image segmentation, single connected component search, and minimum rectangle contour optimization methods, the problem of occlusion recognition of scrap steel on vehicles under full load conditions was solved, achieving high-precision recognition of the main body of the scrap steel.

CN118898834BActive Publication Date: 2026-07-24ANQING NORMAL UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
ANQING NORMAL UNIV
Filing Date
2024-06-21
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

When fully loaded, the characteristics of scrap steel on board are obscured, making it difficult for existing technologies to efficiently identify scrap steel targets inside the truck, resulting in inaccurate assessment accuracy.

Method used

A deep network image segmentation method is used in conjunction with breadth-first search of single connected components, minimum rectangle contour method, and minimum cost translation optimization method to gradually extract and correct the main image of scrap steel, thereby improving the recognition accuracy.

Benefits of technology

By combining deep networks and a two-stage optimization strategy, the problem of fuzzy feature recognition of scrap steel targets under full load conditions is effectively solved, and the recognition accuracy of scrap steel main body is significantly improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118898834B_ABST
    Figure CN118898834B_ABST
Patent Text Reader

Abstract

The application discloses a kind of full load state vehicle-mounted scrap steel automatic identification method, system, equipment and medium, comprising: obtaining the image of vehicle-mounted scrap steel to be identified;Using the image segmentation depth network constructed in advance to the image of vehicle-mounted scrap steel to be identified Segmentation, obtain preliminary target area image;Using the breadth-first search algorithm of single connected domain to the scrap steel main body independent non-evaluation main body in preliminary target area image is excluded, obtain intermediate target area image;Using secondary minimum rectangular contour line method to the scrap steel main body in intermediate target area image is polygonal correction, obtain final target area image;Using minimum cost translation target optimization method extracts the scrap steel main body in final target area image, obtain accurate scrap steel main body image.The application greatly improves the identification precision of scrap steel main body.
Need to check novelty before this filing date? Find Prior Art