The Key Technology Study On The Surface Defects Detection Of Thermal Heavy Rail Based On Machine Vision

The heavy rail plays a decisive role in the national economy production and its quality is an important condition that safeguards the railroad safe transportation. The heavy rail's surface defect is an important attribute that affects the quality of the heavy rail, the thermal heavy rail defect inspection have the quite vital role in the aspects of strengthening the rolling technics management, improving the working condition and the quality of the heavy rail. The machine vision is a method which obtains pictures from the camera by analysis of the computer. For the past decades, the manufacturing industry made the huge progress in the aspects of new material, new craft, new equipment, and so on. Many inspection technologies in the traditional significance have not been able to satisfy the demand of the modern manufacturing industry. The visual inspection technology has the prominent merits of non-contact, fast speed, appropriate precision, strong ability for scene antijamming, and so on which demonstrating the broad application prospect in reality.In this paper, with one large iron and steel enterprise as research background, we have made lots of research on the thermal heavy rail defect image and common detecting methods of it. In view of existing deficiencies of detecting defect through naked eye, this paper has developed a new method for defect detection, which combines the technology of machine vision, image segmentation and SVM recognition method. The main research contents are as follows:â‘ Based on the analysis towards the movement characteristics and inspection technology of the thermal heavy rail, suitable on-line measuring points have been designed; After giving an analysis of surface detection technology at home and abroad, we adopt machine vision technology to collect complete image of the thermal heavy rail directly and non-contactly. Moreover, the hardware and software system have been designed according to the characteristic of the thermal heavy rail surface and technical requirements.â‘¡R esearch the method of real-time image acquisition and high speed transmitting image. The system adopts discretization mode and four independent acquisition units and each one possesses the good interchangeability. This system chooses suitable camera, lens, makes good use of its own infrared light,setting acquisition parameters adjusting the location and imaging angle of the acquisition units reasonablely. It guarantees the speed of image transmission with TCP/IP transmission mode.â‘¢After obtaining the pictures of heavy rail, using the technique of image processing and the algorithm of defect information positioning to reduce ROI and decrease interference. Through studying the categorized method of the defect, we have chosen the SVM-based on the method for the characteristics of few defective samples and irregularity, and training samples repeatedly with the expertise to achieve the goal of differentiating defective kind.â‘£The software scheme chooses Windows operation system 2000 as system platform, VC++ 6.0 and Matlab 7.0 as development platform, and the database technology of SQL Sever is applied in this system. The software system perfectly combine controlling parameters of the camera, real-time image acquisition, image processing, and database management together.Through the above aspects of work, we have carried on overall analysis to the characteristic of the surface defects of the thermal heavy rail, completed a series of systematic studies on the surface defect detection from the theoretical analysis to the complete construction of detection system, and the actual experiments. The on-the-spot experiments testify that the system can effectively recognize the heavy rail surface defects, achieved the anticipated results.