中文摘要：

几何不变量是物体几何结构信息的抽象与概括，在识别中可解决由目标移动产生的旋转或变形等问题。针对现有算法在提取几何不变量时，易造成误差增大和效率不高的情况，提出了一个基于区域面积比的几何不变量构造算法。该算法利用物体图像质心和凸补区域质心构成的直线，对灰度图像的面积区域应用划分策略构建几何不变量的矢量表示形式。在fish及coil-100数据集上的实验表明，算法得到的不变量特征满足仿射不变性，具有良好的区分辨别能力。

英文摘要：

Geometric invariant is the abstraction and generalization of geometrical information from the object, it solves the recognition problems generated by the target movement such as rotation or deformation. To solve the issue of error increase and low efficiency in feature extraction for geometric invariant algorithms, this paper proposed a construction algorithm of geometric invariant based on area ratio. First, the algorithm computed the coordinate positions of a centroid and a convex complement area＇ s centroid of a gray image after binaryzation. Then, it used a division strategy to partition the gray image to some areas according the line passes through these two centroids. Finally, it achieved the area ratio vector of geometric invariant by this algorithm. The experiments in the fish database and eoil-100 of Columbia University show that invariant extracted by the proposed algorithm satisfies affine invariance. Meanwhile, under a certain range of interference （ erase, smear and occlusion） , the invariant features extracted from the object image have discrimination ability.