python利用图像处理方法 实现多目标定位与裁剪（源代码）.zip

import numpy as np import cv2 import copy from PIL import Image # mask # def mask_oprition(mask_path,img_path): # #resize # size = 640 # img = cv2.imread(img_path) # height, width = img.shape[:2] # 得到行和列的长度 # scale=size/height # 得到640和图像高的缩放比例 # height=size # width=int(width*scale) # img = cv2.resize(img, (width, height)) # 读者自行添加mask # mask # mask = cv2.imread(mask_path) # rows,cols,channels = img.shape # gray_back = cv2.cvtColor(mask, cv2.COLOR_BGR2GRAY) # for i in range(rows): # for j in range(cols): # if gray_back[i,j]==0:#0代表黑色的点 # img[i,j]=255#此处替换颜色，为BGR通道 # cv2.imwrite('test_dataset/imgs/masked.jpg',img) #crop def crop_single_picture(img_path): #step1：加载图片，转成灰度图,并用Sobel算子计算x，y方向上的梯度，之后在x方向上减去y方向上的梯度，通过这个减法，我们留下具有高水平梯度和低垂直梯度的图像区域。 image = cv2.imread(img_path) shape = image.shape image_row = shape[0] image_col = shape[1] #显示图片长宽 # print(image_row,image_col) gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) # 得到灰度图 gradX = cv2.Sobel(gray, ddepth=cv2.CV_32F, dx=1, dy=0, ksize=-1) # x方向一阶导数 gradY = cv2.Sobel(gray, ddepth=cv2.CV_32F, dx=0, dy=1, ksize=-1) # y方向一阶导数 # 从x梯度中减去y梯度 gradient = cv2.subtract(gradX, gradY) #显示梯度图 cv2.imshow("gradient1", gradient) gradient = cv2.convertScaleAbs(gradient) # 将图像转回uint8形式 cv2.imshow("gradient2", gradient) #step2：去除图像上的噪声。 blurred = cv2.blur(gradient, (9, 9)) # 滤波，去除噪声 (_, thresh) = cv2.threshold(blurred, 60, 255, cv2.THRESH_BINARY) # 图像二值化，thresh为得到的图像 thresh = cv2.erode(thresh, None, iterations=5) thresh = cv2.dilate(thresh, None, iterations=4) #step3:找出图像所有的轮廓 (cnts, _) = cv2.findContours(thresh.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE) # cnts为得到的轮廓，存放在一个list中 print('轮廓框的个数：{}'.format(len(cnts))) all_box_img = copy.deepcopy(image) # 深复制，跟原图没有关系，一个改变不影响另一个 cv2.drawContours(all_box_img, cnts, -1, (0, 255, 0), 1) # 在all_box_img上画轮廓 count = 0 boxs =[] for i, contour in enumerate(cnts): area = cv2.contourArea(contour) # 计算包围形状的面积 if area < 550: # 读者自行更改：此时包围形状面积小于550的过滤 continue count += 1 rect = cv2.minAreaRect(contour) # 检测轮廓最小外接矩形，得到最小外接矩形的（中心(x,y), (宽,高), 旋转角度） box = np.int0(cv2.boxPoints(rect)) boxs.append(box) # 最后剩下的有用的框 final_box_img = copy.deepcopy(image) print('过滤后轮廓框的个数：{}'.format(len(boxs))) cv2.drawContours(final_box_img, boxs, -1, (0, 255, 0), 1) # 把剩余的框画到原图上 # 裁剪 for i,box in enumerate(boxs): xs = [j[0] for j in boxs[i]] ys = [j[1] for j in boxs[i]] x1 = min(xs) x2 = max(xs) y1 = min(ys) y2 = max(ys) if x1 < 0: x1 = 0 if x2 > image_col: x2 = image_col if y1 < 0: y1 = 0 if y2 > image_row: y2 = image_row print(x1,x2,y1,y2) hight = y2 - y1 width = x2 - x1 cropImg = image[y1:y1+hight, x1:x1+width] cv2.imshow(f"croped{i}", cropImg) cv2.imwrite(f"shiyings/{i}.jpg", cropImg) # 读者自行按照需要显示相应的状态图片 # cv2.imshow("blurred", blurred) # cv2.imshow("gradient", gradient) # cv2.imshow("thresh", thresh) # cv2.imshow("all_box_img", all_box_img) # cv2.imshow("final_box_image", final_box_img) cv2.waitKey(0) if __name__ == '__main__': # 图片路径 img_path = '' # mask路径 #mask_path = 'test_dataset/imgs/in_white_resized.jpg' # 保存路径 path = '' #mask_oprition(mask_path,img_path) crop_single_picture(path)