Canny边缘特征18维描述符在图像拼接中的应用

摘要

为将同一场景中具有重叠区域的序列图像合成为一幅宽视角、高分辨率的图像,基于尺度不变特征变换(SIFT)算法,提出一种结合Canny特征边缘18维特征描述的图像拼接方法.采用SIFT算法提取图像特征点,利用12个圆形区域中的梯度方向累加值、3个同心圆区域中的灰度累加值及3个灰度差分值建立18维特征描述符,保留图像Canny边缘16邻域特征点,选用RANSAC算法进行特征匹配对提纯,计算出不变换矩阵H,并用渐进渐出算法完成图像融合.实验结果表明,与传统SIFT算法相比,该方法对光线变化、旋转、尺度缩放等图像均能取得较好的拼接效果.%In order to gain the wide view angle and high resolution image stitched by the sequence images overlapped in the same scene,based on Scale Invariant Feature Transform (SIFT)algorithm,the image stitching method of 18-dimensional feature description combined with Canny feature edge is proposed.SIFT algorithm is used to extracts feature points of the image 18-dimensional feature descriptor is constructed by using the 12 gradient values in a circular window,3 cumulative gray-scale value and 3 gray-scale difference in a concentric circular window.The features in the 16 neighborhoods of image Canny edge are preserved and the coarse matching points are purified by the RANSAC algorithm.The transformational matrix H is calculated and the image fusion is completed by the slipping into and out algorithm.Experimental results show that compared with the traditional SIFT algorithm,the proposed method has better stitching effect on images with light variation,rotation,and scale scaling.