摘要
针对图像拼接方法中存在的特征提取精度低,以及拼接后存在的拼接裂缝和"GHOST"现象等问题,基于SIFT特征检测的图像拼接.通过采用图像特征点提取和匹配有较强的稳定性和精确度的SIFT特征检测算法,且通过采用较低复杂度的动态规划算法找到最佳缝合线,最后对拼接后的图像通过泊松融合进行平滑处理来完成图像的拼接,并采用自行拍摄的图像进行仿真实验.仿真实验结果表明,基于SIFT特征检测的图像拼接方法具有较高的稳定性和特征提取精度,同时具有较低的特征点提取误差,并对图像拼接中存在的拼接裂缝和"GHOST"现象有很好的抑制作用.
In view of the low precision of feature extraction, the stitching cracks and the ＂GHOST＂ phenomenon existing in the image mosaic method, an image mosaic method based on SIFT feature detection is proposed in this paper. By using the image feature extraction and matching of SIFT feature detection algorithm with stronger robustness and accuracy, and by using the dynamic programming algorithm with low complexity, the best suture line is found, at the end, the stitched image is smoothed by the Poisson fusion to complete image stitching, and the image of their own shooting simulation. Simulation results show that the stability and accuracy of feature extraction method for image mosaic based on feature detection with high SIFT is proposed, also which have the characteristics of low point extraction error, good inhibition effect on splicing cracks existed in image mosaic and the phenomenon of ＂GHOST＂.
出处
《哈尔滨师范大学自然科学学报》
CAS
2017年第2期75-79,共5页
Natural Science Journal of Harbin Normal University
基金
广东高校特色创新类项目(2015KTSCX184)
企业委托项目"图像识别定位软件攻关项目"(2016HKJ042801)
广东省本科高校教学质量与教学改革工程立项建设项目(2015SZL08)
广东省高校优秀青年创新人才培养计划资助项目(2013LYM_0110)
关键词
SIFT算法
图像拼接
动态规划算法
泊松融合
SIFT algorithm
Image stitching
Dynamic programming algorithm
Poisson Fusion
