多尺度最稳定极限区域仿射不变特征  

Affine Invariant Feature of Multi-Scale Maximally Stable Extremal Region

作　　者：罗荣华[1] 闵华清[2] 陈聪[1] 

机构地区：[1]华南理工大学计算机科学与工程学院,广州510006 [2]华南理工大学软件学院,广州510006

出　　处：《计算机研究与发展》2011年第3期508-515,共8页Journal of Computer Research and Development

基　　金：国家自然科学基金项目(60873078;61005061);广东省科技攻关计划基金项目(2009A040300008);中央高校基本科研基金项目(2009ZM0123);深圳市科技攻关计划基金项目(JC200903180740A)

摘　　要：基于局部区域的仿射不变特征被广泛应用于目标识别、场景分类和图像检索.在已经提出的仿射不变局部特征中,最稳定极限区域特征MSER(maxi mally stable extremal region)在多个方面具有优越的性能.但是由于最稳定极限区域特征MSER是从单一尺度图像中提取的,当图像尺度发生较大变化时,图像的模糊会使最稳定极限区域特征的边界发生变化,从而影响特征的稳定性.针对这一问题,通过定义多尺度空间中极限区域的稳定性指标,提出一种在图像空间和尺度空间都最稳定的极限区域特征,并设计了在尺度空间进行极限区域提取的快速算法.同时,针对极限区域可以较好地描述特征轮廓的特点,将局部灰度梯度信息和形状信息相结合设计了一种新的特征描述器.这种特征被称为多尺度最稳定极限区域MMSER(multi-scale maxi mally stable extremal region)特征.实验结果表明,在不同仿射变化条件下,MMSER的稳定性和可识别性均优于MSER,而且其描述器的创建时间约为SIFT描述器的45%.Affine invariant features based on local region are widely used for object recognition,scene classification and image retrieval in the field of computer vision.Among the proposed affine invariant region features,maximally stable extremal region（MSER）has been proven to have attractive properties in several aspects.But since MSER is extracted from one single scale image,MSER will become unstable when the image is blurred due to the change of scale.To solve this problem,an innovational affine invariant feature,which is maximally stable both in image space and scale space,is designed with a criterion to evaluate the stability of extremal regions in scale space.And a fast algorithm is also proposed for extracting extremal regions in the scale space.At the same time,according to the property that the extremal regions can describe the contour of feature fairly well,a new kind of feature descriptor is designed by combining the local gray grads and the shape information.The proposed affine invariant feature with its descriptor is called multi-scale maximally stable extremal region（MMSER）feature.The experimental results prove that MMSER is much more stable and discernable than MSER under different affine transform conditions,and that the computational time of the designed descriptor is about 45% of that of SIFT descriptor.

关 键 词：尺度空间 最稳定极限区域 特征描述器 仿射不变 不变特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]