Python+Opencv实现图像匹配功能（模板匹配）

1、、原理1.1 相似度度量指标 1.2 计算步骤Templa+1)。(w、代码实现2.1 单模板匹配单个标# 相关系数匹配法: cv2.TM_CCOEFFres = cv2.matchTemplate(img, template, cv2.TM_CCOEFF)min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)left_top = max_loc # 左上right_bottom = (left_top0 + w,left_top1 + h) # 右下cv2.rectangle(img, left_top, right_bottom, 2

2、55, 2) # 画出矩形位置plt.subplot(121), plt.imshow(res, cmap=gray)plt.title(Matching Result), plt.xticks(), plt.yticks()plt.subplot(122), plt.imshow(img, cmap=gray)plt.title(Detected Point), plt.xticks(), plt.yticks()plt.show()2.2 单模板匹配多个标import cv2import numpy as npimg_rgb = cv2.imread(mario.jpg)img_gray

3、= cv2.cvtColor(img_rgb, cv2.COLOR_BGR2GRAY)template = cv2.imread(mario_coin.jpg, 0)h, w = template.shape:2res = cv2.matchTemplate(img_gray, template, cv2.TM_CCOEFF_NORMED)threshold = 0.8# 取匹配程度于%80的坐标loc = np.where(res = threshold)返回的坐标值(x,y)是，注意h,w的顺序for pt in zip(*loc:-1):bottom_right = (pt0 + w,p

4、t1 + h)cv2.rectangle(img_rgb, pt, bottom_right, (0, 0, 255), 2)cv2.imwrite(img.jpg,img_rgb)cv2.imshow(img, img_rgb)cv2.waitKey(0)3、算法精度优化NMS BSBDBIoUBBimport cv2import timeimport numpy as npdef py_nms(dets, thresh):Pure Python NMS baseline.#x1、y1、x2、y2、以及score赋值# （x1、y1）（x2、y2）为box的左上和右下标x1 = dets:,

5、 0y1 = dets:, 1x2 = dets:, 2y2 = dets:, 3scores = dets:, 4#每个候选框的积areas = (x2 - x1 + 1) * (y2 - y1 + 1)#order是按照score降序排序的order = scores.argsort():-1# print(order:,order)keep = while order.size 0:i = order0keep.append(i)#计算当前概率最矩形框与其他矩形框的相交框的坐标，会到numpy的broadcast机制，得到的是向量xx1 = np.maximum(x1i, x1order

6、1:)yy1 = np.maximum(y1i, y1order1:)xx2 = np.minimum(x2i, x2order1:)yy2 = np.minimum(y2i, y2order1:)#计算相交框的积,注意矩形框不相交时或算出来会是负数，代替w = np.maximum(0.0, xx2 - xx1 + 1)h = np.maximum(0.0, yy2 - yy1 + 1)inter = w * h#计算重叠度IOU：重叠积/（积1+积重叠积）ovr = inter / (areasi + areasorder1: - inter)#找到重叠度不于阈值的矩形框索引inds =

7、np.where(ovr = template_threshold)#于模板阈值的标坐标score = resres = template_threshold# 于模板阈值的标置信度#将模板数据坐标进处理成左上、右下的格式xmin = np.array(loc1)ymin = np.array(loc0)xmax = xmin+wymax = ymin+hxmin = xmin.reshape(-1,1)#变成列维度xmax = xmax.reshape(-1,1)#变成列维度ymax = ymax.reshape(-1,1)#变成列维度ymin = ymin.reshape(-1,1)#变成

8、列维度score = score.reshape(-1,1)#变成列维度data_hlist = data_hlist.append(xmin)data_hlist.append(ymin)data_hlist.append(xmax)data_hlist.append(ymax)data_hlist.append(score)data_hstack = np.hstack(data_hlist)# 将xmin、ymin、xmax、yamx、scores按照列进拼接thresh = 0.3#NMS的IOU交互阈值keep_dets = py_nms(data_hstack, thresh)pr

9、int(nms time:,time.time() - start_time)# 打印数据处理到nms运时间dets = data_hstackkeep_dets# 最终的nms获得的矩形框return detsif _name_ = _main_:img_rgb = cv2.imread(mario.jpg)#需要检测的图img_gray = cv2.cvtColor(img_rgb, cv2.COLOR_BGR2GRAY)#转化成灰template_img = cv2.imread(mario_coin.jpg, 0)#模板图template_threshold = 0.8#模板置信度dets = template(img_gray,template_img,template_threshold)cou