基于目标规格化与Zernike矩相结合的目标识别算法研究

平面形状的识别是计算机视觉研究的一个重要领域，它也是许多二维或三维视觉处理任务的基本步骤。文章提出了一种基于图像规格化与Zernike矩相结合的目标识别方法。该方法通过把图像进行紧凑化，再结合紧凑化后图像边缘轮廊Zernike矩的计算，使得由于人眼或摄像机的观察视角的不同而引起的物体的各种形变，校正到一个紧凑图像，再通过各高阶Zernike矩的不同特征，进行不同模式的目标识别，实验表明该方法原理简单明了，计算量小，对于特征相关较明显的目标，具有很好的识别效果。

Studies on Object Recognition Based on Combination of Image Compacting and Zernike Moments

This paper proposed a meth od of recognition of planar shapes.It first uses the compacting algorithm bring the image to its most compact form,thus settled the problem of the translation ,scaling,skewing of the image resulting from the change of our viewpoint to the objects,then by computing the Zernike moments of the contour of the compac ting image,it can detect the different characteristics of different kinds of objects and recognize them.This method is based on a simple therom and shows a less computation task,but has a good recognition result.