三维凸面分析法在迷彩伪装检测中的应用

针对边缘检测算法无法检测具有三维凸面性能的伪装目标的缺陷,引入三维凸面检测算子,分析发现,该算子对目标的凸面性能可产生无穷大值响应,但是对噪声相当敏感,从而导致检测结果不准确。由此提出,采用空间域平滑滤波器对图像进行预处理,经预处理后的图像具有良好的抗噪性。通过伪装场景的实际检验证明,改进的方法可有效利用目标具有的三维凸面这一形状特征所表现的灰度差异,成功检测到良好伪装下的迷彩伪装目标。

Application of three-dimensional convex analysis in pattern painting camouflage detection

Edge detection is one of the important means to camouflage detection, but the previous edge detection algorithm can not detect the camouflaged targets with three-dimensional convex properties. Thus the operator of the camouflaged targets detection based on the convexity was proposed, which can get an infinite value as response to the convexity. A smoothing filter of the spatial domain was used to preprocess the images in the camouflaged targets detection because of the operator's intense sensitivity to the noise. By testing of real camouflaged scenes,the improved method was improved to be effective to detect and recoginze the targets under strong camouflage by using the difference of the gray-scale represented by the characteristics of 3D convex shape.