A Segmentation Algorithm for Thermal Paint Image

Thermal paint is widely used for measuring surface temperature of devices in aeronautics and space industry, such as space shuttle´s engine, airplane´s skin etc. The segmentation for the thermal paint´s image is the fundamentals of the digital and automated temperature measuring process. A new method for thermal paint image segmentation is proposed in this paper, which based on color and spatial information. First, PGF algorithm is selected to reduce image´s noises and keep the edge feature of image. Furthermore, using K-Means cluster algorithm, colors in the image are quantized to several representing classes that can be used to differentiate regions in the image. Finally, a region growing and merge method is then used to segment the image. Experimental results demonstrate that good segmentation is obtained by our algorithm, and our algorithm is suitable for segmenting the thermal paint image