基于几何差异的目标识别算法

为降低目标识别算法复杂性且提高其抗噪能力,提出一种基于几何特征差异的目标识别算法。将获取到的目标图片经图像处理后提取轮廓,并以最小周长多边形算法构造目标轮廓的近似多边形;然后根据模板库标准目标做放大或缩小处理后使其面积与模板面积相等；再使用摆放算法使其与模板库图形部分重合；并提出一种改进型双向链表算法求多边形相交部分,通过计算相交部分面积大小达到识别图像的目的。经过仿真实验验证了此方法简单易行,能够快速识别目标。

A New Target Recognition AlgorithmBased on Geometric Difference

In order to reduce the complexity of the target recognition and improve their noise immunity, a geometric-difference based recognition method is presented. First, exact contour of target picture after image processing, and construct approximate polygon outline of target image using a minimum perimeter polygon method. Then zoom in the two polygons according to standard targets in template library to get the same area. And make target contour parts overlap with standard contour using the placement algorithm. After that, we introduced a new algorithm called modified two-way chain table algorithm to get intersection area, and recognized the target according to intersecting area. The simulation experiments show that this method is simple and can quickly identify the target.