复杂自然背景下的人造目标检测

为了解决图像处理中目标检测的问题,提出了一种新的检测方法.该方法根据自然物体和人造物体在纹理上表现出的不同特性,首先将图像划分为若干个子块,然后按各个子块梯度的幅度和方向进行像素编码,最后利用设定的标准检测人造目标.仿真结果表明,新方法可用于不同分辨率的图像,且具有较好的鲁棒性,能够节省存储空间和运算时间,适合于实时处理的场合。

Research on Artificial Objects Detecting in Complicated Natural Environments

For resolving objects detecting problem in digital image processing,a new detecting method is presented based on the fact that the difference of surface texture between artificial objects and natural ones.Firstly,the image is divided into several sub-frames.Secondly,the image pixels are coded according to the amplitude and orientation of their gradient.Lastly,artificial objects are detected by predefined detection criteria.Experimental results show that the proposed method is robust,memory-saving,high speed and suitable to different resolution images.