基于最小矩形包围的图像归一化研究

Image Toward Normalization Study Based on the Smallest Rectangle Surrounded

图像匹配方法要求比对图像灰度、比例和朝向一致。运动目标朝向随机不确定,影响准确匹配识别目标。提出最小矩形包围归一化方法,通过求最小矩形包围时的图像旋转角度,能够将匹配图像朝向归一化为相同的方向。实验表明,提出的方法能够有效解决匹配图像方向不一致问题。

The image matching method requires unanimous image gray, proportion and orientation, and random orientations of moving targets can affect the accuracy of matching. A normalization method based on the smallest rectangle surrounded is raised. When the image rotation angle can be worked out as the surrounded rectangular area is the smallest, the matching images facing in the same direction can be maken. Besides, experimental results show that the proposed method can effectively solve the problem of unanimous orientation when matching the images.