python基于opencv的传统指针式仪表读数

import logging from math import cos, pi, sin import cv2 import numpy as np from sklearn.cluster import KMeans from sklearn.utils import shuffle from kd import get_rad_val from auxiliary.find_roi_FlANN import find_roi from auxiliary.cut_roi import polylines_cut from auxiliary.perspective_roi import perspective_roi import warnings warnings.filterwarnings('ignore', category=FutureWarning) # 旧 # def get_match_rect(template,img,method): # '''获取模板匹配的矩形的左上角和右下角的坐标''' # w, h = template.shape[1],template.shape[0] # res = cv2.matchTemplate(img, template, method) # mn_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res) # # 使用不同的方法，对结果的解释不同 # if method in [cv2.TM_SQDIFF, cv2.TM_SQDIFF_NORMED]: # top_left = min_loc # else: # top_left = max_loc # bottom_right = (top_left[0] + w, top_left[1] + h) # return top_left,bottom_right # 新 def get_match_rect(template, img, method): '''获取模板匹配的矩形的左上角和右下角的坐标''' # template 模版 img 原图 method 匹配方法 # 由于模板的大小和原图片中的大小不一致，所以需要对图片模板大小进行缩放，以准确匹配源图像中的尺寸 max_val_ = 0 max_loc_ = 0 w_ = 0 h_ = 0 count = 0 for dscale in np.linspace(0.1, 2, 100): # 对模板的大小进行缩放，从0.1倍到2倍大小 count += 1 # print(count) template_re = cv2.resize(template, (int(template.shape[1] * dscale), int(template.shape[0] * dscale))) w, h = template_re.shape[1], template_re.shape[0] # 0 为数组长度（高） 1为数组列数宽度（宽） # print("w:{0}h:{1}".format(w,h)) try: res = cv2.matchTemplate(img, template_re, method) / (w * h) # 滑动窗口在源图中寻找模板图，并给出每个位置的匹配相似程度 # print("res:{0}".format(res)) min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res) # 找到像素值的最大和最小值的位置 if max_val > max_val_: max_val_ = max_val max_loc_ = max_loc w_ = w h_ = h except: pass max_loc = max_loc_ w = w_ h = h_ # 使用不同的方法，对结果的解释不同 if method in [cv2.TM_SQDIFF, cv2.TM_SQDIFF_NORMED]: top_left = min_loc else: top_left = max_loc # print("top_left:{0}".format(top_left)) bottom_right = (top_left[0] + w, top_left[1] + h) return top_left, bottom_right def get_circumference(top_left, bottom_right): x1, y1 = top_left x2, y2 = bottom_right w = x2 - x1 h = y2 - y1 long = 2 * (w + h) return long def get_center_point(top_left, bottom_right): '''传入左上角和右下角坐标，获取中心点''' # print("{0}+{1}/2={2}".format(np.array(top_left), np.array(bottom_right), # (np.array(top_left) + np.array(bottom_right)) / 2)) c_x, c_y = ((np.array(top_left) + np.array(bottom_right)) / 2).astype(int) return c_x, c_y def get_circle_field_color(img, center, r, thickness): '''获取中心圆形区域的色值集''' temp = img.copy().astype(np.int) cv2.circle(temp, center, r, -100, thickness=thickness) return img[temp == -100] def v2_by_center_circle(img, colors): '''二值化通过中心圆的颜色集合''' for i in range(img.shape[0]): for j in range(img.shape[1]): a = img[i, j] if a in colors: img[i, j] = 0 else: img[i, j] = 255 def v2_by_k_means(img): '''使用k-means二值化''' original_img = np.array(img, dtype=np.float64) src = original_img.copy() delta_y = int(original_img.shape[0] * (0.4)) # 原始图片中提取类似模板图片的高的0.2倍 （0.2需要修改的参数） delta_x = int(original_img.shape[1] * (0.4)) # 原始图片中提取类似模板图片的宽的0.2倍 （0.2需要修改的参数） original_img = original_img[delta_y:-delta_y, delta_x:-delta_x] # 类模板图片的中心区域，用于k-means的训练样本的区域 # uw=img[delta_y:-delta_y, delta_x:-delta_x] # cv2.imshow("ss",uw) h, w, d = src.shape print('类模板的宽、高、通道数:', w, h, d) dts = min([w, h]) print(dts) r2 = (dts / 3) ** 2 # 距离中心位置的距离平方超过这个位置，则清除 （需要修改的参数） print("{0}:".format(r2)) c_x, c_y = w / 2, h / 2 # 图片的中心位置 a: np.ndarray = original_img[:, :, 0:3].astype(np.uint8) # 获取尺寸(宽度、长度、深度) height, width = original_img.shape[0], original_img.shape[1] depth = 3 # print(depth) image_flattened = np.reshape(original_img, (width * height, depth)) ''' 用K-Means算法进行像素的二值分类学习 ''' image_array_sample = shuffle(image_flattened, random_state=0) estimator = KMeans(n_clusters=2, random_state=0) # k-means评估器 estimator.fit(image_array_sample) # 如果k-means处理效果不佳可以缩小算法的训练样本的区域 ''' 为原始图片的每个像素进行类的分配 ''' src_shape = src.shape new_img_flattened = np.reshape(src, (src_shape[0] * src_shape[1], depth)) # 类模板图展开 cluster_assignments = estimator.predict(new_img_flattened) # 对类模板图的每个像素点进行分类 ''' 建立通过压缩调色板和类分配结果创建压缩后的图片 ''' compressed_palette = estimator.cluster_centers_ # 获取聚类中心，每行为rgb的值，两个聚类点 # print(compressed_palette) a = np.apply_along_axis(func1d=lambda x: np.uint8(compressed_palette[x]), arr=cluster_assignments, axis=0) # 把每个像素点变为到聚类中心点上 img = a.reshape(src_shape[0], src_shape[1], depth) # print(compressed_palette[0, 0]) threshold = (compressed_palette[0, 0] + compressed_palette[1, 0]) / 2 img[img[:, :, 0] > threshold] = 255 img[img[:, :, 0] < threshold] = 0 cv2.imshow('K-Means Binarization', img) cv2.waitKey(1) cv2.destroyAllWindows() for x in range(w): for y in range(h): distance = ((x - c_x) ** 2 + (y - c_y) ** 2) # 距离图片中心位置较远的区域，清空处理 if distance > r2: pass img[y, x] = (255, 255, 255) cv2.imshow('sd', img) cv2.waitKey(1) cv2.destroyAllWindows() return img def dilate(img): '''将Kmeans二值化后的图像进行膨胀处理''' kernel = np.ones((3, 3), np.uint8) dilation = cv2.dilate(img, kernel, iterations=1) return dilation def get_pointer_rad(img): '''获取角度''' shape = img.shape c_y, c_x, depth = int(shape[0] *0.51), int(shape[1] *0.52), shape[2] # 寻找指针的旋转中心位置 （0.75是需要修改的参数） # c_y, c_x, depth = shape[0]//2, shape[1]//2, shape[2] # 寻找指针的旋转中心位置 （0.75是需要修改的参数） x1 = c_x + c_x * 0.8 src = img.copy() freq_list = [] for i in range(0, 361): # 拟合指针的角度范围，水平向右为0度，逆时针为正方向（使用自己的数据集，需要自定义的参数） # x = (x1 - c_x) * cos(i * pi / 180) + c_x # y = (x1 - c_x) * sin(i * pi / 180) + c_y x = (x1 - c_x) * cos(np.radians(i)) + c_x y = (x1 - c_x) * sin(np.radians(i)) + c_y temp = src.copy() cv2.line(temp, (c_x, c_y), (int(x), int(y)), (0, 0, 255), thickness=2) # 绘制直线 t1 = img.copy() t1[temp[:, :, 2] == 255] = 255 c = img[temp[:, :, 2] == 255] cv2.imshow("ssd",t1) points = c[c == False] freq_list.append((len(points), i)) cv2.imshow('d'