Thermopaint Color Image Segmentation And Temperature Recognition System

Temperature field measurement is an important subject in both practice and theory. It is the essential point of research to a special system of thermopaint image recognition base on image processing, which could automate and digitize the process of analysis, overcome the defect of artificial, increase the efficiency of the measuring accuracy.The system is non-contact, flexibility, accuracy, efficiency and so on. Firstly, aiming at the noisy problem that is produced by the surroundings and the hardware, a filtering method is proposed base on threshold, then the image is quantized according to the average color moment, and HSI color space is chosen base on handmade consistent with subject vision. This paper mainly research two key technologies which lie in the thermopaint color image segmentation algorithm and temperature recognition algorithm:(1)Color image segmentation methods suiting to the thermopaint are proposed which Compared and analysis the currently problem of color image segmentation. One algorithm is based on labeling region growing, and each region will be labeled; another one is based on the combination with wavelet and fuzzy clustering, and clustering algorithm use the wavelet energy feature and the image color feature as feature vector.(2)After image segmentation and feature extraction, this paper carries on spline interpolation to color value and temperature value corresponding relations in view of color-temperature characteristic, and produces a color-temperature characteristic curve, so the temperature is recognized according to this curve.The results showed that this system with the segmentation and recognition algorithm of this paper can work steadily and obtain accurate detection results, which can never be matched by the manual detection.