摘要
微创外科手术中的图像特征点快速匹配,可使计算机具备图像实时识别能力,提高手术成功率.但由于在手术中所运用的图像匹配算法具有计算量大、耗时长等缺点,提出一种基于SURF的图像特征点快速匹配算法.首先对图像采用SURF算法提取特征点,然后通过Hear小波变换确定特征点的主方向和特征点的描述子,并使用改进的最近邻搜索算法进行特征点匹配,最终根据实际需要选取相似度最高的前50～100对匹配点进行对比实验.实验结果表明,该算法鲁棒性强、速度快、匹配准确性高,且在医学图像处理中具有较大的应用价值.
Image feature points fast matching in the minimally invasive surgery offers the computer real-time im- age recognition capabilities ＂and improves the success rate. However, this algorithm is computationally intensive and time-consuming. A SURF-based fast image feature points matching algorithm is proposed. First, the image feature points are extracted using the SURF algorithm, then the main direction and the descriptors of the feature point are determined via Hear wavelet transform and the improved nearest neighbor search algorithm is used for matching fea- ture points, and finally the first 50 to 100 points with the highest similarity are compared selected according to actual needs for matching experiments. Experimental results show that the algorithm has good robustness, high speed and high matching accuracy, and is of value in medical image processing.
出处
《电子科技》
2014年第5期145-148,共4页
Electronic Science and Technology
基金
湖南省教育厅科研基金资助项目(09B004)
关键词
图像匹配
特征点
最近邻搜索算法
image matching
features point
SURF
best bin first algorithm
