提出一种基于兴趣点特征匹配(IPFM)的图像自动镶嵌技术。首先利用混合的角和边检测器获取待镶嵌图像中的兴趣点,通过采用一种双向最大相关系数匹配的方法提取出初始特征点对,然后用空间坐标距判别准则剔除伪特征对,最后利用正确匹配特征点对实现图像的镶嵌。此外,在图像融合技术中引入了域值的概念,从而解决了重叠区域图像平滑过渡问题。实验表明:该方法能够快速准确地提取出两幅图像间的对应特征点,大大降低了误匹配的概率;在图像间存在小的旋转量情形下,仍能有效地实现图像的平滑镶嵌。
An automatic image mosaic technique based on interest points feature matching (IPFM) was proposed. Firstly the combined corner and edge detector is used to gain the interest points,and a kind of bidirectional greatest correlative coefficient match law is used to extract the initial feature points pair, then the false feature points pair is rejected by space coordinate distance criterion,finally the correct match feature points pair is used to realize image stitching; In addition,the field value concept is introduced in the image fusion technology to solve the smooth transition problem through overlap region. The experimental results indicated that this method can fast and accurately extract the corresponding feature points between two images and reduce the probability greatly which matches by mistake. Under the existing slight rotator situation, this method still could effectively realize the image smoothly stitching.