光学技术, 2019, 45 (1): 78, 网络出版: 2019-04-16  

结合几何代数改进的SIFT桥梁裂缝图像拼接算法

Improved SIFT algorithm for bridge crack image mosaic combining geometric algebra
作者单位
1 上海理工大学  机械工程学院, 上海  200093
2 上海应用技术大学  城市建设与安全工程学院, 上海 201418
摘要
针对桥梁裂缝图像精度要求高,拼接质量受原图像亮度变化大、噪声干扰严重和对比度低的影响,提出了一种结合几何代数改进的SIFT桥梁裂缝图像的新型拼接算法。对SIFT算法进行了两方面的改进: 一是通过几何代数空间的表示形式提取了待拼接图像的色度图像,克服了SIFT算法中色度信息丢失的不足; 二是改进了SIFT算法对灰度图像建立尺度空间的方法,构建了新的可适用于多光谱图像的高斯滤波和卷积运算,确定了尺度空间。通过几何代数DoG空间检测特征点并进行预匹配。使用改进的RANSAC算法对匹配结果进行修正,完成了图像之间的精确拼接。实验结果表明,所提算法的性能优于SIFT算法,提取的特征点对数量提高了近10%; 拼接过程中未产生位错现象,最终拼接结果满足桥梁裂缝图像的精度要求。
Abstract
Aiming at the high accuracy of bridge crack image and the quality of the mosaic image which is subject to large brightness transformation, serious noise interference and low contrast, a novel SIFT bridge crack image mosaic algorithm combined with geometrical algebra (GA) is presented. The SIFT algorithm is improved in two aspects. The chrominance image of the image to be spliced is extracted through the representation of geometric algebraic space, and the deficiency of chrominance information loss in the SIFT algorithm is overcome. The other hand is that the SIFT algorithm is improved in establish scale space for gray image, and a new Gaussian filtering and convolution operation applicable to multispectral image is constructed, and the scale space is determined. The feature points are detected through the GA-DoG space and the pre-match between images is completed. The improved RANSAC algorithm is used to correct the matching results and complete the precise stitching between images. Experimental results show that the performance of this algorithm is better than SIFT algorithm, and the number of extracted feature point pairs is increased by nearly 10%.There is no dislocation occurred during splicing. The final splicing result satisfies the accuracy requirements of bridge crack images.

王艳, 李宗学, 丁文胜, 沈晓宇. 结合几何代数改进的SIFT桥梁裂缝图像拼接算法[J]. 光学技术, 2019, 45(1): 78. WANG Yan, LI Zongxue, DING Wensheng, SHEN Xiaoyu. Improved SIFT algorithm for bridge crack image mosaic combining geometric algebra[J]. Optical Technique, 2019, 45(1): 78.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!