object detection[NMS][非极大抑制]

非极大抑制，是在对象检测中用的较为频繁的方法，当在一个对象区域，框出了很多框，那么如下图：



上图来自这里

目的就是为了在这些框中找到最适合的那个框，主要就是通过迭代的形式，不断的以最大得分的框去与其他框做iou操作，并过滤那些iou较大（即交集较大）的框

IOU也是一种Tanimoto测量方法[见模式识别，希腊，书609页]

按照github上R-CNN的matlab代码，改成py的，具体如下：

def iou(xminNp,yminNp,xmaxNp,ymaxNp,areas,lastInd,beforeInd,threshold):

#将lastInd指向的box，与之前的所有存活的box指向坐标做比较
xminNpTmp = np.maximum(xminNp[lastInd], xminNp[beforeInd])
yminNpTmp = np.maximum(yminNp[lastInd], yminNp[beforeInd])
xmaxNpTmp = np.maximum(xmaxNp[lastInd], xmaxNp[beforeInd])
ymaxNpTmp = np.maximum(ymaxNp[lastInd], ymaxNp[beforeInd])

#计算lastInd指向的box，与存活box交集的，所有width，height
w = np.maximum(0.0,xmaxNpTmp-xminNpTmp)
h = np.maximum(0.0,ymaxNpTmp-yminNpTmp)
#计算存活box与last指向box的交集面积
inter = w*h
iouValue = inter/(areas[beforeInd]+areas[lastInd]-inter)

indexOutput = [item[0] for item in zip(beforeInd,iouValue) if item[1] <= threshold ]
return indexOutput

def nms(boxes,threshold):
'''
boxes:n by 5的矩阵，n表示box个数，每一行分别为[xmin,ymin,xmax,ymax,score]
'''
assert isinstance(boxes,numpy.ndarray),'boxes must numpy object'
assert boxes.shape[1] == 5,'the column Dimension should be 5'

xminNp = boxes[:,0]
yminNp = boxes[:,1]
xmaxNp = boxes[:,2]
ymaxNp = boxes[:,3]
scores = boxes[:,4]
#计算每个box的面积
areas = (xmaxNp-xminNp)*(ymaxNp-yminNp)
#对每个box的得分按升序排序
scoresSorted = sorted(list(enumerate(scores)),key = lambda item:item[1])
#提取排序后数据的原索引
index = [ item[0] for item in scoresSorted ]
pick = []
while index:
#将当前index中最后一个加入pick
lastInd = index[-1]
pick.append(lastInd)
#计算最后一个box与之前所有box的iou
index = iou(xminNp,yminNp,xmaxNp,ymaxNp,areas,lastInd,index[:-1],threshold)

return pick

if __name__ == '__main__':

nms(boxes,threshold)

参考资料：

[] 非极大抑制。http://www.cnblogs.com/liekkas0626/p/5219244.html

[] Theodoridis.S.,.Koutroumbas.K..Pattern.Recognition,.4ed,.AP,.2009