基于FCM聚类的示温漆图像分割算法

根据示温漆(TIP)图像特征及相关实验结果，提出了一种基于模糊C均值(FCM)聚类的示温漆图像分割算法。该算法以FCM聚类算法为核心，应用双边滤波器平滑图像，提高图像分割质量。通过颜色空间转换，提高图像聚类的迭代效率。实验表明:该算法增强了示温漆温度判读需要的特征信息，可以提取人眼无法识别的等温线，同时消除了等温线崎岖等现象，测温误差为±10℃。算法应用效果良好，满足工程要求，已用于示温漆自动判读软件开发。 

Segmentation algorithm for temperature indicating paint image based on FCM clustering

According to the image features and the related experimental results, a temperature indicating paint(TIP) image segmentation algorithm based on fuzzy C-means (FCM) clustering was proposed. Taking the FCM clustering algorithm as the core, the algorithm employed the bilateral filter to smooth the image to improve the segmentation quality. The color space conversion was used to improve the iterative efficiency of the image clustering. Experiments showed that the algorithm can enhance the feature information needed in the process of the interpretation of the TIP. The algorithm can extract isotherms that cant be recognized by naked eye and the temperature measurement error was ±10℃. The algorithm has good application effect and meets the engineering requirements, so it has been applied to the development of automatic interpretation software for TIP.