Quality Inspection Online System Of Spring Based On MV

A lot of springs were used in the laying of track parts. In order to ensure the quality of spring, on-line detection of parts mechanical size is needed. Because of the parts on-line at a high temperature state, manual inspection is inefficient and there is also a certain risk.Machine vision detection technology has many advantages, such as non-contact, online time, speed, precision appropriate and strong anti-interference abilities. It is adapted to the progress and development of modern manufacturing requirements. It actually has broad application prospects. This paper designs a line for spring quality inspection system, combining with machine vision and image processing technology. In order to achieve the system, this paper has done research in the following areas: 1. design a reasonable image acquisition system based on the full investigation of the on-site environment and the actual testing requirements; 2.focus on studying camera calibration technique, to ensure the accuracy of size measurement; 3. establish a suitable image processing algorithms with a large number of experiments, and also achieve a spring-line dimensional inspection and the expected detection accuracy.A suitable system structure on the detection system is essential. Through field investigation and laboratory environment, this paper makes a detailed design and scientific analysis of the hardware structure of the machine vision quality inspection system. Based on the construction of the basic framework, this paper introduces system structure and principle. And analysis of the errors.In order to meet the requirements of precision and speed, this article discusses the calibration techniques deeply, analysis of the current calibration method, made a brief introduction on ant colony algorithm and neural network. Compared with other calibration techniques, neural networks with ant colony optimization for BP camera calibration have increased in accuracy and robustness.The article outlines the threshold segmentation and edge detection algorithm, selects the most suitable system of programs. In the image acquisition system, features for spring in the industrial cameras with filter to increase substantially improve the quality of the image acquisition, and ultimately successfully extracted the feature points. Finally, the article makes a brief introduction on the system interface and function, and measures the quality of the spring, and the measurement accuracy of 0.1mm which is met the design requirements, and discusses the results of measurements and analyzes errors. Measurement system has been working in the factory, and successfully detected a million fragments.Studies show that the detection system used in the program and methods are feasible and correct, and indicators of the performance reached the expected requirements. This research project has some theoretical and practical significance.