一种自适应遥感图象小目标检测新方法

针对遥感图象普遍存在的小目标,提出了一种新的自适应检测方法.该方法基于图象局部扭曲度和峭度所具有的非对称性和拖尾特性,自适应地确定具有小目标的感兴趣区域(ROI),并在该区域内,结合能量特征实现图象目标的分割.针对分割分的二值图象仍具有的干扰噪声和虚假目标问题,进行了二值数学形态滤波,并结合目标的面积特征实现了小目标的精确检测.计算机模拟仿真结果证明了所提出方法的有效性.

A new adaptive detecting method for small man-make objects in the remote sensing images

Aimed at the small man-made objects in the remote sensing images,a new adaptive detecting method is proposed.Based on the local skewness and kurtosis,which is a measure of the asymmetry and impulsiveness of the distribution in the image,the regions of interest (ROI)are decided adaptively.Combined with the energy feature in ROI,the candidates of the small man-made objects are segmented.On the segmented binary images,the morphological filters and area features are applied to reduce the noise and false objects.The computer simulation results show that the new method is an effective means for small man-made objects detection.