摘要
提出了一种基于尺度不变特征变换(SIFT)和小波变换的图像拼接算法,以提高室外复杂场景的图像拼接质量.利用SIFT算法提取基准图像(待匹配图像)和后续图像(与基准图像进行匹配的图像)的特征点,确定特征点的位置、尺度与方向;利用128维向量对特征点进行描述;利用最近邻法完成两幅图像特征点的匹配,确定重合区域;利用基于小波变换的多分辨率方法完成对图像的拼接.实验结果表明,该方法对亮度差异较大的图像拼接效果良好,适宜于室外复杂环境的图像拼接.
An algorithm of image stitch based on scale invariance feature transform（SIFT） and wavelet transform is proposed to improve the stitch effect of significantly different luminance images. Features are first detected between two images using SIFT algorithm. After identified locations and scales, features are described by a 128-dimensional feature vector. Feature is matched based on the nearest neighbor method in order to identify overlap regions. Images are blended by wavelet transform. Experimental results showed that the presented method is effective for significant different luminance images adopted from outdoor environments.
出处
《北京理工大学学报》
EI
CAS
CSCD
北大核心
2009年第5期423-426,共4页
Transactions of Beijing Institute of Technology
基金
国家自然科学基金与微软亚洲研究院联合资助项目(60673198)
国家“八六三”计划项目(2008AA01Z303)
关键词
尺度不变特征变换
小波变换
图像拼接
特征点匹配
图像融合
scale invariance feature transform（SIFT）
wavelet transform
image stitch
feature match
image blending
