基于主目标区域形状不变矩的图像匹配算法

Image Matching Algorithm Owing to Host Target Area Form Invariable Regulation

基于不变矩图像匹配虽然克服了性能的不稳定性，在相对平移、旋转尺度变换的鲁棒性等方面具有一定的优势．但由于其计算公式复杂，匹配耗时长，实时性差。针对此提出了一种新的基于主目标区域的形状的图像匹配算法。此算法先取出待匹配图像的主目标区域，通过提取主目标区域的7个矩不变量图像特征，用图像特征向量的归一化欧氏距离表示图像的相似度，并用花朵等图像作为示例图像进行验证，实验证明该算法具有一定的优越性。

Based on the same moment although the image matching the performance to overcome the instability in the relative translation, rotation-scale transformation of the robustness of the areas have certain advantages, but because of its complicated formula, matching a long time-consuming, real-time worse. For this proposed a new main target area based on the shape of the image matching algorithm. Algorithm to match images out to be the main target area, the main target area through the extraction of seven moment variable image feature, with images of the eigenvector Normalization of Euclidean distance that the similarity of images, images of flowers and used as a sample image to verify the experiment proved that the method has certain advantages.