这是在论文中，用matlab中神经网络bp方法训练红外图像象素值和温度之间对应关系的程序-This the thesis, using Matlab neural network training method bp infrared image pixel and temperature relationships between the procedures

图象 处 理和模式识别中经常根据物体的形状来识别物 体或对物体进行分类。形状特征可以分为区域特征和边界特 征两大类，矩不变量是最基本的形状特征。Hu提出的矩不变 量是图象区域内部细节的描述，是一种区域特征。它是建立 在对一个区域内部灰度值的统计分析基础上的，需要目标区 域的所有象素参与运算-Image processing and pattern recognition are often based on the shape of objects to identify o

基于噪声检测的中值滤波，先通过邻域计算检测是否为噪声，再进行开关中值滤波，对于被判断为噪声的像素不再参与其他像素的中值滤波，能够有效滤除噪声并且保留细节较好-Noise Detection Based on median filtering, first calculated by testing whether the neighborhood noise, then the switching median filter for noise pixels are judged to no l

基于熵率的超像素分割方法Matlab代码(Super pixel segmentation method based on entropy rate Matlab code)