改进ORB图像匹配联合最佳缝合线在数字媒体视频图像拼接中的应用
Application of improved ORB image matching combined with optimal stitching in digital media video image stitching
郭蕾1
作者信息
摘要
视频监控设备极大程度丰富了数字化信息时代下的图像数据类型和数量,考虑到设备视角对真实场景信息展示的局限性以及传统视频拼接技术所表现出的低质量问题.研究提出从视频图像匹配和融合两方面入手,通过对匹配算法与种子区域生长算法进行改进,引入混合粒子群算法、向量相似度准则、泊松方程等方面保证拼接水平.实验结果表明,改进匹配算法所表现出的正确匹配率均在96％以上,且最佳缝合线下的最大峰值信噪比(23.28)和信息熵(7.758)明显优于其他算法,图像拼接准确率超过95％,表现出较好的主客观评价效果.该方法对于全景图像在大视野、大尺度的应用研究具有参考价值.
Abstract
Video surveillance equipment greatly enriches the types and quantities of image data in the digital information age,considering the limitations of device perspectives on displaying real scene information and the low quality issues exhibited by tradition-al video stitching techniques.The research proposes to start with video image matching and fusion,and improve the matching algo-rithm and seed region growth algorithm by introducing hybrid particle swarm optimization algorithm,vector similarity criterion,Pois-son equation,and other aspects to ensure the stitching level.The experimental results indicate that the optimization matching algo-rithm exhibits a correct matching rate of over 96％,and the maximum peak signal-to-noise ratio(23.28)and information entropy(7.758)under the optimal stitching line are significantly better than other algorithms.The image stitching accuracy exceeds 95％,demonstrating good subjective and objective evaluation effects.This method has reference value for the application research of pano-ramic images in large field of view and scale.
关键词
ORB算法/最佳缝合线/混合粒子群算法/视频图像/泊松方程Key words
ORB algorithm/best suture line/hybrid particle swarm optimization algorithm/video images/poisson equation引用本文复制引用
基金项目
大学生创新创业训练计划项目(S202210722078)
出版年
2024
NETL
NSTL
万方数据