基于视觉的目标识别中改进SIFT算法研究
|
摘要
为解决SIFT特征匹配算法计算量大、运算速度慢等问题,在详细分析了原有经典算法的基础上,对其进行改进:一方面从降低运算复杂度的角度考虑,通过对每个像素的梯度模值和梯度方向进行高斯加权,将原有128维的特征描述符降低至24维;另一方面对特征向量的搜索方法进行改进,在原有BBF搜索方法的基础上,引入每一维度的数据与节点之间的关系来限定搜索范围,减少搜索次数,提高算法的搜索速度,进而减少算法整体的运行时间。最后,通过实验验证算法在运算速度上的提升,同时对匹配的准确度进行分析,在保证算法准确度的基础上提升运算速度。
Because the computational complexity and slow operation exist in SIFT,this paper analyzes and improves the original classic algorithm. To reduces the computational complexity,Gaussian weighting is performed for each pixel gradient model and gradient direction,the feature descriptor is reduced from 128 to 24 dimensions,then,the search range is limited by introducing the relationship between the data of each dimension and the node and the run time of the algorithm is reduced. At the end of this paper,the validity of the algorithm is verified in terms of accuracy and speed.
Because the computational complexity and slow operation exist in SIFT,this paper analyzes and improves the original classic algorithm. To reduces the computational complexity,Gaussian weighting is performed for each pixel gradient model and gradient direction,the feature descriptor is reduced from 128 to 24 dimensions,then,the search range is limited by introducing the relationship between the data of each dimension and the node and the run time of the algorithm is reduced. At the end of this paper,the validity of the algorithm is verified in terms of accuracy and speed.
引文
[1]王殿君.双目视觉在移动机器人定位中的应用[J].中国机械工程,2013,24(9):1155-1158.
[2]林琳.机器人双目视觉定位技术研究[D].西安:西安电子科技大学,2009.
[3]D. Marr,T. Poggio. A computational theory of human stereo vision[J]. Proceedings of the Royal Society of London,1979,204:301-328.
[4]陈兵旗.视觉技术与应用及应用实例详解[M].北京:化学工业出版社,2014
[5]王民,刘伟光.基于改进SIFT特征的双目图像匹配算法[J].计算机工程与应用,2013,49(2):203-206.
[6] Lowe. G. Distinctive image features from scale-invariant key points[J]. International Journal of Computer Vision,2004,60(2):91-110.
[7]Yan K,Sukthankar R. PCA-SIFT:A more distinctive representation for local image descriptors[C]. Proceedings of the 2004IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2004(2):506-513.
[8]Mikola. A performance evaluation of local descriptors[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2005,27(10):1615-1630.
[9]李丹,孙海涛,王海莉.一种改进的SIFT图像立体匹配算法[J].西南交通大学学报,2015,50(3):490-496.
[10]Beis J S,Lowe D G. Shape indexing using approximate nearestneighbor search in high-dimensional spaces[C]. Proceedings of the CVPR,San Juan,1997:1000-1006.
[11]赵岩,陈月,王世刚.结合投影误差校正的快速SIFT图像拼接[J].光学精密工程,2017,6(25):1645-1651.
[2]林琳.机器人双目视觉定位技术研究[D].西安:西安电子科技大学,2009.
[3]D. Marr,T. Poggio. A computational theory of human stereo vision[J]. Proceedings of the Royal Society of London,1979,204:301-328.
[4]陈兵旗.视觉技术与应用及应用实例详解[M].北京:化学工业出版社,2014
[5]王民,刘伟光.基于改进SIFT特征的双目图像匹配算法[J].计算机工程与应用,2013,49(2):203-206.
[6] Lowe. G. Distinctive image features from scale-invariant key points[J]. International Journal of Computer Vision,2004,60(2):91-110.
[7]Yan K,Sukthankar R. PCA-SIFT:A more distinctive representation for local image descriptors[C]. Proceedings of the 2004IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2004(2):506-513.
[8]Mikola. A performance evaluation of local descriptors[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2005,27(10):1615-1630.
[9]李丹,孙海涛,王海莉.一种改进的SIFT图像立体匹配算法[J].西南交通大学学报,2015,50(3):490-496.
[10]Beis J S,Lowe D G. Shape indexing using approximate nearestneighbor search in high-dimensional spaces[C]. Proceedings of the CVPR,San Juan,1997:1000-1006.
[11]赵岩,陈月,王世刚.结合投影误差校正的快速SIFT图像拼接[J].光学精密工程,2017,6(25):1645-1651.