| 引用本文: | 涂国勇,李壮,周韶斌,等.Gabor二进制编码异源图像匹配方法.[J].国防科技大学学报,2015,37(5):175-179.[点击复制] |
| TU Guoyong,LI Zhuang,ZHOU Shaobin,et al.Gabor binary encoding for multi-sensor image matching[J].Journal of National University of Defense Technology,2015,37(5):175-179[点击复制] |
|
| |
|
|
| 本文已被：浏览 10311次 下载 6784次 |
| Gabor二进制编码异源图像匹配方法 |
| 涂国勇1,2, 李壮1,2, 周韶斌1, 李伟建1, 于友合3 |
|
(1. 中国酒泉卫星发射中心， 甘肃 酒泉 732750;2.
2. 国防科技大学 航天科学与工程学院， 湖南 长沙 410073;3. 中国人民解放军96326部队， 湖南 怀化 418000)
|
| 摘要: |
| 异源图像匹配是图像处理领域尚未解决的问题。其中，合成孔径雷达图像与光学图像差异较大，用现有方法匹配通常难以得到满意结果。针对这个问题，提出一种基于Gabor编码的异源图像匹配方法：选取一组Gabor滤波器，分别对大图和小图进行Gabor卷积；采用池化方法对卷积结果进行压缩表示；对池化结果二值化并转换为二进制表示得到Gabor二进制编码特征；采用二进制位操作计算实时图与基准图对应窗口特征的相似性，相似性最大值对应图像匹配结果。本方法采用二进制对图像进行描述，减少了计算量，同时也更好地描述了异源图像间的共性特征。实验结果表明，本方法具有较高的匹配概率，计算时间少于现有方法。 |
| 关键词: 图像匹配 异源图像 Gabor滤波器 二进制编码 特征池化 |
| DOI：10.11887/j.cn.201505027 |
| 投稿日期：2014-10-26 |
| 基金项目:国家自然科学基金资助项目（61402489） |
|
| Gabor binary encoding for multi-sensor image matching |
| TU Guoyong1,2, LI Zhuang1,2, ZHOU Shaobin1, LI Weijian1, YU Youhe3 |
|
(1. Jiuquan Satellite Launch Center, Jiuquan 732750, China;2.
2. College of Aerospace Science and Engineering, National University of Defense Technology, Changsha 410073, China;3. The PLA Unit 96326, Huaihua 418000, China)
|
| Abstract: |
| Multi-sensor image matching is a challenging problem in image process field. As synthetic aperture radar images and optical images have significant differences, most existing methods cannot achieve satisfied matching result. To respond to this issue, a new multi-sensor image matching method based on Gabor binary encoding was presented: the big and small input images were first convoluted respectively by a group of Gabor filters; the compressed representation was executed on the convolution result by using pooling method; the binarization of pooling results was conducted and it was transformed into binary code to create Gabor binary encoding features; the similarities of corresponding window features between real-time images and reference images were calculated by using bit manipulation and the maximum value indicated the matching result. This method describes images by binary representation, so the computation complexity is much lower than that of the traditional method, while the common characters are better revealed. Experimental results show that the proposed method has much higher matching rate and require much lower computation time than those of the existing methods. |
| Keywords: image matching multi-sensor image Gabor filters binary coding feature pooling |
|
|
|
|
|
