摘要
针对当前方法在进行红外数字图像局部序列拼接过程中存在效率不高、拼接图像质量较差等问题,提出一种基于最佳路径与亮度权重函数相结合的红外数字图像局部序列无损拼接方法,通过利用SIFT算法对经过直方图均衡化处理的红外数字图像在尺度空间中进行极值点检测、局部特征点的精确定位、局部特征点的方向分配、生成特征点描述,完成红外数字图像SIFT特征点提取;在图像SIFT特征点提取基础上,运用相似度度量方法中的局部相关系数对提取到的红外数字图像SIFT特征点对进行匹配,引入最佳路径与亮度权重函数对红外数字图像匹配重叠区域的像素亮度差异进行平滑,同时消除重影,实现了红外数字图像局部序列无损拼接。实验结果表明,所提方法不仅大大提高了拼接效率,而且保证了红外数字图像拼接的高质量,减小了失真和错位。
Aiming at the problems of low efficiency and poor stitching image quality in the process of infrared image local sequence stitching in the current method,a method of lossless stitching of infrared image local sequence based on optimal path and luminance weight function is proposed. The SIFT algorithm is used to perform extreme point detection,localization of local feature points,local feature point direction assignment,and feature point descriptor generation in the scale space of the infrared digital image processed by the histogram equalization to complete the extraction of SIFT feature points of the infrared digital image. Based on the SIFT feature points extraction,the SIFT feature point pairs of the extracted infrared digital images are matched by the local correlation coefficient in the similarity measurement method,and the best path and luminance weight function are introduced to match the infrared digital image matching area. The pixel brightness difference of the area is smoothed,ghosting is eliminated at the same time,and the local sequence of the infrared digital image is non-destructively stitched. The experimental results show that the proposed method not only greatly improves the splicing efficiency,but also guarantees the high quality of infrared digital image mosaic and reduces distortion and misalignment.
作者
邸臻炜
朱肖颖
黄筱佟
DI Zhenwei;ZHU Xiaoying;HUANG Xiaotong(School of Information&Electronic Engineering,Wuzhou University,Wuzhou Guangxi 543002,China;Guangxi Colleges and Universities Key Laboratory of Image Processing and IntelligentInformation System,Wuzhou University,Wuzhou Guangxi 543002,China)
出处
《激光杂志》
北大核心
2018年第12期47-51,共5页
Laser Journal
基金
广西自然科学基金资助项目(No.2015GXNSFBA139262)
梧州学院校级科研项目(No.2014C035)
关键词
红外数字图像
局部序列
无损
拼接方法
infrared digital image
local sequence
lossless
stitching method
