形状是人类通过视觉识别物体时所需的关键信息之一，它不随周围环境的变化而

Shape is one of the key information that humans need to recognize objects visually. It does not change with changes in the surrounding environment and is relatively stable information of objects. <br>The extraction algorithms for shape features can be roughly divided into two categories: contour-based and region-based. The contour-based representation method is to describe the contour of the surrounding target area. Common methods include Fourier descriptors, chain codes, curvature scale space, and boundary moments. Since the shape of the area is treated as a whole, All pixels in the area can be effectively used, so it is less affected by noise and shape changes. Shape-based region feature description methods include eccentricity, dispersion, region invariant moment, angular radius transformation, geometric invariant moment and so on.

形状是人类通过视觉识别物体时所需的关键信息之一，它不随周围环境的变化而变化，是物体较为稳定的信息。<br>对于形状特征的提取算法大致可以分为两类：基于轮廓和基于区域。基于轮廓的表示方法是对所包围的目标区域的轮廓进行描述，常用的方法有傅立叶描述符、链码、曲率尺度空间、边界矩等，由于是将区域内的形状当作一个整体来对待，可以有效地利用区域内的所有像素，因此受噪声和形状变化的影响比较小。基于形状的区域特征描述方法有偏心率、离散度、区域不变矩、角半径变换、几何不变矩等等。

Shape is one of the key information for human beings to recognize objects by vision. It does not change with the change of surrounding environment, and it is a relatively stable information of objects.<br>There are two kinds of shape feature extraction algorithms: contour based and region based. The representation method based on contour is to describe the contour of the surrounding target area. The commonly used methods are Fourier descriptor, chain code, curvature scale space, boundary moment, etc. because the shape of the area is treated as a whole, all the pixels in the area can be effectively used, so the influence of noise and shape change is relatively small. The shape based region feature description methods include eccentricity, discreteness, region invariant moment, angle radius transformation, geometric invariant moment and so on.<br>