基于色彩空间变换的目标检测算法研究

Study on Target Detection Algorithm Based on Color Spatial Conversion

对高光谱遥感图像的目标进行检测,通常采用异常检测算法.异常检测算法常在检测到目标的同时出现较多大噪声点和非目标异常点,为解决这一问题提出了一种基于HSV色彩空间的目标检测算法.利用背景和目标的光谱曲线特征,找到三个差异最大的特征波段子集进行平均,再进行假彩色合成并正变换至HSV色彩空间,以H分量为阈值进行目标检测.对仿真数据和真实高光谱数据进行了实验,实验结果证明了本算法的有效性.

Anomaly detection algorithm is usually used to detect targets in hyperspectral images. Many big noise points and non-target anomaly points appear when targets were detected by anomaly detection algorithm. In order to solve this problem, a target detection algorithm based on color spatial conversion is presented. Average the three band subsets which has the most obvious variance were found out by the spectrum curves features of background and targets. Then these three band subsets were used to get false color image and transform to HSV color spatial. H component is used to detect targets as threshold. Simulation data and real hyperspectral data experiments showed that this method is effective.