The Research Of An Image Processing Method In Infrared Image Metal Fatigue Limit Detection

Recent years, people are paying more and more attention to mental fatigue limit, so it’s very important to find an exact, effective and fast method of metal fatigue limit detection. Because of the excellent features of wide measurement range, fast measurement speed, non contact and night vision of infrared camera, an infrared image metal fatigue limit detection method was presented which can not only reflect the stress distribution of workpieces but also forecast some bugs in early time. But there are still some problems which baffle its practical applications, for example, it’s difficult to detect valuable signals in high-noise condition and to get quantitative calculation of fatigue limit. These are the problem restricting the application of infrared camera.This paper deeply analyzed the method of metal fatigue limit detection in infrared image, its content included:(1) Researching the input data of infrared camera, designing a file format which was named PTW and was convenient to carry out the theory of thermo-elasticimetry and dissipative energy detection. Analyzing the file format of BMP file and designing Visual C++.net programs to make PTW file transform into BMP file;(2) Considering the features of signals and noises in metal fatigue limit detection, signal model which was adapt to thermo-elasticimetry and dissipative energy detection was established;(3) Carrying out the algorithm of thermo-elasticimetry detection in infrared image based on getting load signals, and settling the problem that detecting thermo-elasticimetry signals are difficult in high-noise condition. Further more, some new VC++.NET programs were designed on infrared images of Piston-Bolt and Crank-Piston-Pole. At the same time, the de-noising effect of the new algorithm was compared with that of medium filter and mean filter. Finally the de-noising effect of the new algorithm and the classical algorithm was estimated by two indexes-signal to noise ratio and mean square error;(4) Solving the problem which is preferable to detect dissipative energy in low SNR condition based on preamble;(5) Designing a Layered-K-means clustering algorithm after analyzing the disadvantage and advantage of Layered clustering and K-means clustering, the infrared images are Pseudo-color processed by the new algorithms and the definition of de-noised images was also improved by it;(6) Developing a infrared image metal fatigue limit detection system software, designing the interface, meanwhile the algorithms and function of (1)-(5) were carried out. At last the analysis and algorithms of detection method is proved correct and effective through experiment results.