【摘要】 针对二值商标图像的特点,提出了一种综合边界曲率特征和图像分块熵矩阵特征的检索算法。首先,根据微分几何中曲率的定义,计算图像形状边界上每一点的曲率,并统计得到曲率直方图作为边界特征。然后,在图像分块的基础上,计算每一分块子图像的信息熵,得到熵矩阵,求该矩阵的奇异值作为区域特征。最后,综合这两个特征进行检索。实验表明,边界和区域特征综合使用较之单一特征有着更好的检索效果,并具有较好的几何不变性。更多还原

【Abstract】 An image retrieval method based on curvature feature and entropy-matrix feature for binary trademark images was proposed.First,according to the definition of curvature in Differential Geometry,every curvature of the fringe point was computed,and curvature-histogram was obtained as boundary feature.Second,the image was parted into several blocks,and then entropy for every block was counted.The entropy-matrix consisted of these entropy of every block.Last,images were retrieved by the two features.Experimental result shows that using the feature of boundary and region is better than using the single one,and it has better anti-geometric distortion effect.更多还原