摘要
针对图像采集和测量时需要将多幅图像拼接成一幅大图像的问题,提出了一种图像拼接算法,利用特征区域块分割技术来实现图像的拼接.用灰阶Sobel算子通过引入衰减因子对图像进行边缘检测得到不失真的灰阶边缘图,然后将灰阶边缘图进行三次样条插值处理,使特征区域块边缘的定位达到亚像素级,提高了图像边缘检测的精度,有利于图像的高精度拼接.实践结果表明,该算法简单易行,定位较准确,拼接效果好,能有效地克服图像质量的影响,当图像灰度值相差较大并且图像具有旋转时,该算法也能实现对图像无缝平滑拼接.
A merging algorithm for images which are captured and need to be measured is proposed, which introduces the segmentation of feature regions into image merging. The edges of an image is detected by the gray Sobel operator with an attenuation factor. A gray edge image is obtained without distortion. Then the gray edge image can be processed using cubic spline interpolation. The location precision of feature regions is improved to subpixel. It improves the precision of edge detection and is beneficial to merging images with high precision. The experimental result shows that the algorithm is effective, simple and easy to use for images merging correctly. When the images have a biggish difference in graylevel and rotation, the algorithm can effectively overcome the influence in image quality and make merging images seamless and smooth.
出处
《西安电子科技大学学报》
EI
CAS
CSCD
北大核心
2002年第6期768-771,共4页
Journal of Xidian University
