Add files via upload

211-谷海涛/第二周/img2gray.py

@@ -0,0 +1,73 @@
+import cv2
+import numpy as np
+from skimage.color import rgb2gray
+import matplotlib.pyplot as plt
+
+#灰度化
+#1、手工实现
+img = cv2.imread('1.png')    #读入图片，绝对路径或相对路径
+height, width, channels = img.shape
+img_gray = np.zeros([height, width], img.dtype)  #使用np.zeros生成宽、高与图片像素一致的全部是0的矩阵，准备存储生成后的值
+for i in range(height):                          #遍历img矩阵，取出每个像素的BRG值
+    for j in range(width):
+        img_gray[i, j] = int(0.11 * img[i, j][0] + 0.59 * img[i, j][1] + 0.3 * img[i, j][2])   #Gray = 0.3R+0.59G+0.11B，cv2读取的顺序是BGR
+cv2.imwrite('img_gray.png', img_gray)    #保存
+
+#2、调用cv接口
+img = cv2.imread('1.png')
+img_gray_2 = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)     #直接调用接口即可，cv2.cvtColor，注意cv2读取的是BGR顺序
+cv2.imwrite('img_gray_2.png', img_gray_2)
+
+#3、调用skimage的rgb2gray接口
+img = plt.imread('1.png')    #plt读取之后对矩阵做了归一化，就是所有值都除以255，读取的顺序是RGB
+img_gray_3 = rgb2gray(img)   #from skimage.color import rgb2gray
+plt.imsave('img_gray_3.png', img_gray_3, cmap='gray')
+
+#二值化
+#1、手工实现-cv2
+img = cv2.imread('1.png')
+height, width = img.shape[:2]
+img_gray = np.zeros([height, width], img.dtype)      #通过灰度化实现二值化
+img_binary = np.zeros([height, width], img.dtype)    #构建初始矩阵
+for i in range(height):
+    for j in range(width):
+        img_gray[i, j] = int(0.11 * img[i, j][0] + 0.59 * img[i, j][1] + 0.3 * img[i, j][2])   #灰度化
+        img_bin = img_gray[i, j]/255    #归一化
+        if img_bin >= 0.5:
+            img_binary[i, j] = 255      #二值赋值
+        else:
+            img_binary[i, j] = 0
+cv2.imwrite('img_binary.png', img_binary)
+
+#2、手工实现-plt
+img = plt.imread('1.png')
+height, width = img.shape[0:2]
+img_binary_2 = np.zeros([height, width], img.dtype)
+for i in range(height):
+    for j in range(width):
+        if img_gray_3[i, j] >= 0.5:          #使用上面已经灰度化的img_gray_3，由于已经归一化，所以直接使用即可
+            img_binary_2[i, j] = 1
+        else:
+            img_binary_2[i, j] = 0
+plt.imsave('img_binary_2.png', img_binary_2, cmap='gray')
+
+#3.使用numpy
+img = plt.imread('1.png')
+img_gray = rgb2gray(img)
+img_binary_3 = np.where(img_gray >= 0.5, 1, 0)      #书写简单
+plt.imsave('img_binary_3.png', img_binary_3, cmap='gray')
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+

211-谷海涛/第二周/nearest_insert.py

@@ -0,0 +1,23 @@
+import cv2
+import numpy as np
+
+
+def nearest_insert(img, new_size):
+    ori_height, ori_width, channels = img.shape    #获取原图片的高度、宽度及通道数
+    new_height, new_width = new_size
+    new_img = np.zeros((new_height, new_width, channels), img.dtype)   #创建新图片大小的矩阵
+    h_ratio = new_height/ori_height     #计算放大比例
+    w_ratio = new_width/ori_width
+    for i in range(new_height):
+        for j in range(new_width):
+            x = int(i/h_ratio + 0.5)   #使用round进行四舍五入时候需注意边界处理
+            y = int(j/w_ratio + 0.5)
+            new_img[i, j] = img[x, y]
+
+    return new_img
+
+
+img = cv2.imread('1.png')
+size= [1000, 1000]
+nearest_insert_img = nearest_insert(img, size)
+cv2.imwrite('nearest_insert.png', nearest_insert_img)