| 引用本文： |
-
胡宇,许慧娜,王少军,曹东.陆海一体数字孪生智慧渔场平台中的视频图像拼接方法[J].广西科学院学报,2024,40(4):379-388. [点击复制]
- HU Yu,XU Huina,WANG Shaojun,CAO Dong.Video Image Stitching Method for Digital Twin Intelligent Fisheries Integrating Land and Sea[J].Journal of Guangxi Academy of Sciences,2024,40(4):379-388. [点击复制]
|
|
| 摘要: |
| 对于智慧渔场管理中面对较大空间范围的全局实时监测需求，监控视频图像拼接可以解决由于监控视频零散造成的应对不及时甚至误判等问题。针对智慧渔场视频图像拼接过程中图像配准耗时长、特征匹配率低、图像特征纹理弱等问题，本文设计一种BRISK(Binary Robust Invariant Scalable Keypoints)+GMS(Grid-based Motion Statistics)组合式图像配准算法。在BRISK快速提取特征点的基础上，使用结合双向匹配策略的GMS-RANSAC (Random Sample Consensus)算法筛选优秀特征点对，使得后续的图像融合更加平滑自然。为了解决图像融合过程出现的拼接缝、重影等问题，本文设计一种结合改进渐入渐出融合算法的最佳缝合线算法，优化最佳缝合线算法中的能量函数，引入HSV (Hue,Saturation,Value)色彩空间中的色彩饱和度代替色差强度，使得图像拼接在最小色差区域进行，同时优化渐入渐出融合算法中的权值函数，从而消除缝合线以及重影现象，使图像融合中的图像过渡更加平滑自然。实验验证了图像配准阶段和图像融合阶段算法在视频图像拼接系统中的可行性和实用性。 |
| 关键词: 图像处理 图像拼接 图像配准 图像融合 智慧渔场 数模融合 |
| DOI：10.13657/j.cnki.gxkxyxb.20241226.002 |
| 投稿时间：2024-04-01修订日期：2024-04-28 |
| 基金项目:国家重点研发计划专项(2022YFD2401200)资助。 |
|
| Video Image Stitching Method for Digital Twin Intelligent Fisheries Integrating Land and Sea |
|
HU Yu1,2, XU Huina1,2, WANG Shaojun2, CAO Dong2
|
| (1.College of Computer Science and Artificial Intelligence, Zhengzhou University, Zhengzhou, Henan, 450001, China;2.Zhongke Xingtu Jinneng (Nanjing) Technology Co., Ltd., Nanjing, Jiangsu, 211100, China) |
| Abstract: |
| For the demand of global real-time monitoring in a large spatial range in the management of intelligent fishing grounds,surveillance video image stitching can solve the problems of untimely response or even misjudgment caused by scattered surveillance video. Aiming at the problems of long time-consuming image registration,low feature matching rate and weak image feature texture in the process of video image mosaic of intelligent fishing ground,a Binary Robust Invariant Scalable Keypoints (BRISK)+Grid-based Motion Statistics (GMS) combined image registration algorithm is proposed. On the basis of BRISK's fast extraction of feature points,GMS-RANSAC (Random Sample Consensus) algorithm combined with bidirectional matching strategy is used to screen out excellent feature point pairs,which makes the subsequent image fusion smoother and more natural. In order to solve the problems of stitching seams and ghosting in the image fusion process,an optimal stitching algorithm combined with an improved fade-in and fade-out algorithm is designed in this article to optimize the energy function in the optimal stitching algorithm. The color saturation in the HSV (Hue,Saturation,Value) color space is introduced to replace the color difference intensity,so that the image stitching is performed in the minimum color difference region. At the same time,the weight function in the fade-in and fade-out fusion algorithm is optimized to eliminate the stitching line and ghosting phenomenon,so that the image transition in image fusion is smoother and more natural. and the weight function in the fade-in and fade-out fusion algorithm is optimized to eliminate the stitching lines and ghosting,so that the image transition in the image fusion is smoother and more natural. Experiments verify the feasibility and practicability of the image registration phase and the image fusion phase algorithm in the video image stitching system. |
| Key words: image processing image stitching image alignment image fusion intelligent fishery data-model integration |
