#coding:utf-8'''图像缩放'''import numpy as npimport cv2image = cv2.imread('datas/l1.jpg')# 基于等隔提取图像缩放def scale_simple(image,kx,ky): # 计算缩放后图像的分辨率 rows = int(np.round(np.abs(image.shape[0] * kx))) cols = int(np.round(np.abs(image.shape[1] * ky))) dist = None if len(image.shape) == 3 and image.shape[2] >= 3: dist = np.zeros((rows,cols,image.shape[2]),image.dtype) else: dist = np.zeros((rows,cols),image.dtype) for y in range(rows): for x in range(cols): new_y = int((y + 1) / ky + 0.5) - 1 new_x = int((x + 1) / kx + 0.5) - 1 dist[y,x] = image[new_y,new_x] return dist# 基于区域子块提取图像缩放def area_average(image,left_point,right_point): temp1,temp2,temp3 = 0,0,0 # 计算区域子块像素点个数 #npix = (right_point['x'] - left_point['x'] + 1) * (right_point['y'] - left_point['y'] + 1) # 对区域子块各个通道对像素值求和 for i in range(right_point['x'] - left_point['x']): i += left_point['x'] for j in range(right_point['y'] - left_point['y']): j += right_point['y'] temp1 += image[i,j,0] temp2 += image[i,j,1] temp3 += image[i,j,2] vec3b = [temp1,temp2,temp3] return vec3bdef scale_area(image, kx,ky): rows = int(np.round(np.abs(image.shape[0] * kx))) cols = int(np.round(np.abs(image.shape[1] * ky))) dist = np.zeros((rows,cols,image.shape[2]),image.dtype) left_row_coord = 0 left_col_coord = 0 for i in range(rows): x = int((i + 1) / kx + 0.5) - 1 for j in range(cols): y = int((j + 1) / ky + 0.5) - 1 left_point = {'x':left_row_coord,'y':left_col_coord} right_point = {'x':x,'y':y} pixel = area_average(image,left_point,right_point) dist[i,j] = pixel left_col_coord = y + 1 left_col_coord = 0 left_row_coord = x + 1 return distresult = scale_simple(image,0.5,0.5)result2 = scale_area(image,0.5,0.5)cv2.imshow('src',image)cv2.imshow('scale-simple',result)cv2.imshow('scale-area',result2)cv2.waitKey()cv2.destroyAllWindows() 