SIFT,SURF,ORB,FAST 特征提取算法比较

---

图像处理的基础就是要进行特征点的提取，feature(interest points) detect 的方法也在不断的进步，边检测，角点检测，直线检测，圆检测，SIFT特征点检测，同时描述符也在发展，为了匹配的高效，逐渐从高维特征向量到二进制向量…下面做一个简单的罗列，并调用OpenCV API看看效果！
承接上一篇文章。

Feature Detection Methods List:

• Canny Edge Detect, A Computational Approach to Edge Detection, 1986. The Canny edge detector is an edge detection operator that uses a multi-stage algorithm to detect a wide range of edges in images.
• Harris, A combined corner and edge detector, 1988. considering the differential of the corner score with respect to direction directly.
• GFTT，Good Features to Track,1994, Determines strong corners on an image.
• Matas-2000, Robust Detection of Lines Using the Progressive Probabilistic Hough Transform. 霍夫变换检测直线.
• SIFT,Distinctive Image Features from Scale-Invariant Keypoints,2004, invariant to image translation, scaling, and rotation, partially invariant to illumination changes and robust to local geometric distortion. 128-dim(512B).
• SURF, Speeded Up Robust Features,2006,受SIFT启发，比SIFT快，健壮. 64-dim(256B).
• FAST ,Machine Learning for High-speed Corner Detection, 2006,wiki. Very fast, not robust to high level noise.
• ORB, ORB: an efficient alternative to SIFT or SURF,2011，基于FAST和BRIEF，比SIFT快两个数量级，可作为SIFT的替代(a fusion of FAST keypoint detector and BRIEF descriptor). 32B binary descriptor.
• BRISK，BRISK: Binary Robust Invariant Scalable Keypoints, 2011. 64B binary descriptor.
• STAR，Censure: Center surround extremas for realtime feature detection and matching,2008,引用次数不高.scale-invariant center-surround detector (CENSURE) that claims to outperform other detectors and is capable of real-time implementation.
• MSER，Robust Wide Baseline Stereo from Maximally Stable Extremal Regions, 2002, 斑点检测(blob detection).

特征点提取算法比较(image dataset(27 pictures)):

ImageNOSIFTSURFORBFASTSTARBRISK0241441265001197871515381429581295001676311661861234044784500161918161445316392802500716620369941510148449729562238334215105728309500720
065619118729516125825178273352470650056715438165403374267011558276294899752350012780473129910197942125001067686414981135993294500663070121631682877923661953131884241350011681548268314250950555001809716712898159177477350072248428881633323217500205021381261217544666115001655368319591845926033500706542161926650945996133565832020872786500745922360721258236515001214772015302225094237500148905071113231236454550064734107182413112606500429319949125237387500657122132269681418488660945343Time Cost21.5217.40.970.252.342.14

上面是通过一些图片集测的单单是feature detect 的时间，接下来通过一对图片看看feature detect和compute feature descriptor总共花费的时间开销（秒）：

Image pairSIFTSURFORBFAST(SURF)eiffel-1,13.jpg2.773.220.110.22

可以看到计算descriptor的开销还是很大的，这里仅仅两张图片，所以主要开始是计算描述符，提取是很快的。

下面通过通过两张图片来看这几个算法匹配的效果，1639-1311-697表示图片1，2分别提取了1639，1311个keypoints，其中匹配的有697个。

Image pairSIFTSURFORBFAST(SURF)BRISKeiffel-1,13.jpg1639/1311/6972802/2606/1243500/500/2511196/1105/586607/491/287

---

Canny Edge Detection效果：

Find line segments by probabilistic Hough transform：

Harris Corner Detection：

SIFT match效果图：

SURF match效果图：

ORB match效果图：

BRISK match效果图：

---

代码在这里

---

参考：
Canny Edge Detector Example
Feature Detection-Canny, HoughLinesP
Harris corner detector Example
BRIEF (Binary Robust Independent Elementary Features)
ORB (Oriented FAST and Rotated BRIEF)