摘要
将马氏距离矢量作为不变矩用于图像中的物体识别 ,并就马氏距离计算过程中物体部件划分问题进行了研究 ,提出了一种基于物体图像灰度的部件点集划分新方法。实验结果表明 ,马氏距离矢量方法应用于图像物体识别 ,对待识别物体在视场中的平移和旋转具有良好的适应性 ,识别误差小 ,所提出的新的点集划分方法可直接根据物体图像的灰度进行点集划分 ,克服了拓扑关系方法不易实现自动划分的缺点 ,实现了基于马氏距离方法图像识别过程的自动化。
This paper applies mahalanobis distance vector as the invariable in recognizing the object in the image. The problem of object division in the process of computing mahalanobis distance is discussed. Based on the gray scale of object image, a new division method is brought forward. Results of experiments proved that this method applied in object recognition has good adaptability and small error when the object moves and rotates in the sight. The new method brought by this paper leads to divide the point set of object image directly according to its gray scale. This method gets rid of the difficulty of the realization of division based on topology, and realizes the automation of image recognition based on mahalanobis distance vector.
出处
《浙江工业大学学报》
CAS
2001年第4期357-359,共3页
Journal of Zhejiang University of Technology
