Research On The Vision Inspection System Of Three-way Catalytic Honey-comb Ceramic

Quality control is an effective method to enhance the enterprise competitiveness, during the production process. In recent years, three-way catalysts are widely used in automobile exhaust emissions. Honeycomb ceramic, installed in the middle of the catalyst, is one of the effectively used carriers. However, it is tough workload for ceramic inspection, due to the current large market volume of honeycomb ceramics, different kind of cross section of the ceramic(e.g., circular, voal or triangular), compulsory inspection of the shape dimension(i.e.,roundness and height) before market saling, defect, the number of holes in per square millimeter and blocked holes. Nowadays, majority of the domestic companies still use the original detection method, sampling by calipers. The certain disadvantages of this basic method are inefficiency, bias, and high cost. Furthermore, it cannot handle the counting of the total number of the holes in per square millimeter. Hence, it has not only great positive impacts to the economy and society but also contributions to alleviate or solve the current serious haze problem, to develop one kind of automatic honeycomb ceramics inspection system and replace the manual inspection method.Because the image detection system with high precision, high speed, high repeatability,non-destructive testing and other advantages, so this paper designed honeycomb ceramic quality detection system based on machine vision technology. This paper described the current situation and prospect of the development of image processing technique,and also it pointed out the times and forward-looking. Then the overall design of quality honeycomb ceramic-line detection system is proposed. Camera selection principles, optical system, light design and installation are analyzed in detail in this paper; It also introduced the conversion between the imaging model of the camera and projection principle. Nonlinear model,proposed by Zhang Z.Y were applied to calibrate the camera. This calibration was served as the foundation for the dimension inspection of the ceramics. Depending on the analysis of the main sources of the image noise, a new hybrid noise filtering algorithm is effective and practical. Finally, as for the edge-detection of the ceramics, “Canny operator” was applied to roughly find out the edge points firstly, Zernike moments method then was applied to position the subpixel edge. The experiments showed that this algorithm can extract a precise edge profile for the online detection system. Meanwhile, with this developed ceramic inspection system, detecting time of the honeycomb ceramic’s maximum and minimum diameter can bedramactively decreased from 15 seconds(manual inspection) to 2 seconds. The detection accuracy can be improved from 0.1mm to0.01 mm, and the error on counting the holes can be decreased to ±4 from ±20. Finally, the sources the inspection error were analyzed in this thesis.