Research on Key Technologies of Online Industrial Intelligent Inspection System Based on Computer Vision
Innovate online computer vision inspection to empower the industry with high-precision and low-cost intelligent manufacturing.
Type
Tags
Applicable industry
Applications
Key innovations
The innovation of this project lies in: proposing nonlinear scale spatio-temporal feature detection to enhance the robustness of target recognition. Introduce an improved compressed sensing algorithm to achieve long-term real-time tracking. It is the first in China to extract three-dimensional information from monocular images based on multi-scale wavelet clues to solve the problem of three-dimensional acquisition in industrial sites.
Potential economic benefits
The computer vision inspection system replaces manual inspection, significantly reduces industrial production costs and improves efficiency and accuracy. Its low cost and high flexibility have expanded the scope of application, has produced significant economic benefits in many enterprises, and promoted industrial intelligence upgrades.
Potential climate benefits
The system significantly reduces product defects and scrapping rates in industrial production through high-precision, real-time online detection. This directly reduces raw material consumption, rework energy consumption and transportation, thereby reducing carbon emissions throughout the production life cycle.
Solution supplier
View more
Shanghai University
Shanghai University is a comprehensive research university in Shanghai, committed to cultivating high-quality innovative talents and serving national and regional economic and social development.
Shanghai，China
Solution details
- Field:
This project belongs to the field of information technology and is a technology applied by computer vision technology in the fields of industrial production and manufacturing and industrial testing. - Main contents:
In the field of industrial manufacturing and production, the concept of "Industry 4.0" has been widely proposed abroad. In the domestic industrial field, most companies still use traditional manual testing methods.
Computer vision is a science that studies how to empower computers to intelligently perceive image data obtained by image sensors. Applying machine vision technology to the industrial field, especially in online intelligent detection, has the characteristics of large information volume, high precision, continuity, high cost efficiency, flexibility, non-contact, wide spectral response range, automatic processing, and easy integration of design information and processing control information and convenient and flexible data management functions.
In view of the relatively poor flexibility and high cost of existing computer vision inspection systems, which require a strong knowledge background, and have fewer bottlenecks in large-scale mature applications, a comprehensive research on the key technologies of online industrial intelligence implemented by Shanghai University was carried out on computer vision-based inspection systems. The main contents include: target object recognition technology, long-term real-time tracking technology, three-dimensional reconstruction technology, image restoration technology during transmission, vision system development and application platform, etc. Based on this key technology, we have innovatively developed a computer vision development platform for online industrial intelligent inspection, and implemented multiple online intelligent inspection application systems for the automobile manufacturing industry, steel industry, tobacco industry and other fields, and carried out Large-scale application. This project has obtained 4 invention patents, applied for 2 new invention patents, and published 16 academic papers. - Innovations and characteristics:
- Target recognition is the foundation of industrial site detection and control. This project proposes a nonlinear scale spatio-temporal feature detection method for the first time in this field, which is highly robust to detecting target position movements, scale changes, illumination and occlusion.
- In view of the problem that in the industrial environment, the target to be detected is easy to be lost during the tracking process, a long-time real-time target tracking algorithm based on improved compressed sensing is introduced for the first time. By integrating tracking and detection, it effectively solves the problem during the tracking process. The target is lost.
- In order to solve the problem of difficulty in obtaining three-dimensional information in industrial sites, a depth extraction algorithm based on multi-scale wavelet clues is innovatively used to directly extract relevant three-dimensional information from two-dimensional images acquired by monocular cameras. The use of this technology is the first in China.
- Aiming at the problem of low-bit-rate wireless video transmission in industrial environments, the optical flow equation is combined with the spatial correlation equation of motion vectors in the application site, and a simple and fast iterative approximation algorithm and the traditional optical flow field based full-frame restoration algorithm are presented, which improves the restoration ability of video images in wireless video transmission.
- Research and develop a customizable computer vision software development system platform based on multiple platforms with the goal of online industrial intelligent detection. Compared with other existing domestic system platforms, this platform is more scalable, has lower cost, and has a number of application examples.
- Application promotion:
At present, this project has been applied and promoted in dozens of units such as Shanghai Transportation Group Company, Shanghai Xintong Mechanical and Electrical Engineering Company, Xuzhou Weighing Machine Factory Company, and China Tobacco Corporation, producing significant economic benefits. The implementation of this project has greatly narrowed the gap between my country and foreign countries in related fields, paved the way for the reliable practical application of computer vision systems, has extremely important value and significance for the industrialization of related fields, and provides an effective technology for the industrialization and modernization of Shanghai City and the construction of "Industry 4.0."
Last updated
01:01:04, Nov 08, 2025
Information contributed by
See original page on
Report