基于大视差图像中目标物体的拼接

Image Stitching of the Target Object Based on Large Parallax Images

近年来,随着计算机技术和多媒体技术的快速发展,全景拼接技术的应用也越来越广泛。而针对于大视差图像中目标物体的拼接技术应用和研究有一定的局限性。大视差图像中目标物体的拼接,指一系列纵向或横向连续垂直拍摄的二维图像中,目标物体与背景视差较大,在拼接时只关注目标物体的拼接而非整幅图像,一般应用于航拍建筑物体的拼接或高度测量。为了获得目标物体较好的拼接效果,采用等距拍摄,以及利用视差图与特征点匹配集平行线选取技术,获取目标物体本身的特征点匹配集,进而完成目标物体的横向或纵向拼接。实验结果表明,该方法能够提高大视差图像中目标物体的拼接效果,并能有效满足大视差图像中目标物体的拼接。

In recent years, with the rapid development of computer technology and multimedia technology, the panorama stitching techniques ＇ applications are increasingly used. But for splicing technology applications and research on the target object in a large parallax images have some limitations. Stitching the target object of large parallax images, referring to a series of two-dimensional image vertically or laterally continuous vertical shooting, the target object and background have larger parallax, when splicing images only focus the target object rather than the entire image, generally applied to aerial architectural objects stitching or height measurement. In order to obtain a better target object mosaic effect, uses isometric shooting, and takes the use of the disparity map and feature points matching set of parallel lines select technology, gets the target object itself feature point matching set, and then completes the horizontal or vertical mosaic of the target object. Experimental results show that this method can improve the mosaic effect large parallax image of the target object, and can effectively meet the stitching large parallax image of the target object.