首页 | 官方网站   微博 | 高级检索  
     

纹理抑制的光照不均图像配准算法研究
引用本文:刘天,曾亮.纹理抑制的光照不均图像配准算法研究[J].计算机工程与应用,2020,56(23):180-188.
作者姓名:刘天  曾亮
作者单位:国防科技大学 计算机学院,长沙 410073
摘    要:针对KAZE算法对光照不均图像特征点提取效果不佳的问题,提出了一种基于纹理抑制改进后的KAZE图像配准算法。改进算法的主要流程如下:将纹理抑制算法嵌入到非线性扩散滤波方程中,以实现对图像更好的光照估计;对光照估计后图像的亮度分量进行自适应Gamma校正;利用改进的KAZE算法对图像进行配准。实验结果表明,改进算法相较于SIFT、SURF、KAZE算法的平均正确匹配率分别提高了48.5个百分点、22.1个百分点和20.1个百分点,查全率提高了26个百分点、5个百分点和5.5个百分点,所提算法能有效地降低误匹配,并且广泛地应用到多种处理场景中。

关 键 词:图像配准  特征检测  特征匹配  KAZE算法  纹理抑制  

Illumination Uneven Image Registration Algorithm Research Using Surround Suppression
LIU Tian,ZENG Liang.Illumination Uneven Image Registration Algorithm Research Using Surround Suppression[J].Computer Engineering and Applications,2020,56(23):180-188.
Authors:LIU Tian  ZENG Liang
Affiliation:College of Computer, National University of Defence Technology, Changsha 410073, China
Abstract:To solve the problem that the KAZE algorithm does not extract the feature points of the uneven illumination image, an improved KAZE image registration algorithm based on surround suppression is proposed. The main process of the improved algorithm is as follows: Firstly, the surround suppression algorithm is embedded into the nonlinear diffusion equation to achieve better illumination estimation of the image. Secondly, adaptive Gamma correction is applied to the luminance component of the image after illumination estimation. Finally, the image is registered using the improved KAZE algorithm. The experimental results show that the average correct matching rate of the improved algorithm is 48.5 percentage points, 22.1 percentage points, and 20.1 percentage points higher than that of SIFT, SURF, and KAZE algorithms respectively and the recall is increased by 26 percentage points, 5 percentage points, and 5.5 percentage points. The proposed algorithm can effectively reduce mismatches and is widely used in a variety of processing scenarios.
Keywords:image registration  feature detection  feature matching  KAZE algorithm  surround suppression  
本文献已被 万方数据 等数据库收录!
点击此处可从《计算机工程与应用》浏览原始摘要信息
点击此处可从《计算机工程与应用》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号