Spaces:
Sleeping
Sleeping
| import matplotlib.pyplot as plt | |
| import matplotlib.colors as mcolors | |
| import numpy as np | |
| import cv2 | |
| def individual_channel_image(img_arr, channel= 'r', ax=None): | |
| img_arr = img_arr[:,:,0:3] | |
| if channel in ['r','red','Red']: | |
| plot_arr = img_arr[:,:,2] | |
| channel_name = 'Red' | |
| cmap = 'Reds' | |
| if channel in ['g','green','Green']: | |
| plot_arr = img_arr[:,:,1] | |
| channel_name = 'Green' | |
| cmap = 'Greens' | |
| if channel in ['b','blue','Blue']: | |
| plot_arr = img_arr[:,:,0] | |
| channel_name = 'Blue' | |
| cmap = 'Blues' | |
| if channel not in ['r','red','Red','g','green','Green','b','blue','Blue']: | |
| plot_arr = img_arr | |
| channel_name = 'Original' | |
| if ax is None: | |
| if channel_name == 'Original': | |
| plt.imshow(cv2.cvtColor(img,cv2.COLOR_BGR2RGB)) | |
| else: | |
| plt.imshow(plot_arr) | |
| plt.colorbar(orientation= 'vertical', shrink = 0.7, pad = 0.01) | |
| plt.title('Image in the {} channel'.format(channel_name)) | |
| plt.show() | |
| if ax is not None: | |
| if channel_name == 'Original': | |
| ax.imshow(cv2.cvtColor(img_arr,cv2.COLOR_BGR2RGB)) | |
| else: | |
| plot = ax.imshow(plot_arr, cmap = cmap) | |
| plt.colorbar(plot, orientation= 'vertical', ax = ax) | |
| # plt.colorbar(orientiation= 'vertical', shrink = 0.7, pad = 0.1) | |
| ax.set_title('Image in the {} channel'.format(channel_name)) | |
| def channel_distribution_plotter(img_array): | |
| img_array = img_array[:,:,:3] #Not considering the A channel, if it's a RGBA image. | |
| plt.subplot(2,2,1) | |
| plt.hist(img_array[:,:,2].ravel(),bins=256,color='red'); | |
| plt.title("Red Channel") | |
| plt.subplot(2,2,2) | |
| plt.hist(img_array[:,:,1].ravel(),bins=256,color='green'); | |
| plt.title("Green Channel") | |
| plt.subplot(2,2,3) | |
| plt.hist(img_array[:,:,1].ravel(),bins=256,color='blue'); | |
| plt.title("Blue Channel") | |
| plt.subplot(2,2,4) | |
| plt.imshow(cv2.cvtColor(img_array,cv2.COLOR_BGR2RGB)) | |
| plt.title("Original Image") | |
| plt.suptitle("Pixel values distribution in each channel\nx-axis: pixel values, y-axis: number of pixels") | |
| plt.tight_layout() | |
| plt.show() | |
| def which_channel_dominates(img_arr, original_image_plot = 'yes', original_image_opacity = 0.3, channel_opacity = 0.7): | |
| cmap = mcolors.ListedColormap(['red', 'green', 'blue', 'white', 'black', 'gray']) | |
| img_arr = img_arr[:,:,:3] | |
| red_channel = img_arr[:,:,2] | |
| green_channel = img_arr[:,:,1] | |
| blue_channel = img_arr[:,:,0] | |
| print(np.max(red_channel), np.max(green_channel), np.max(blue_channel)) | |
| if original_image_plot == 'yes': | |
| plt.imshow(cv2.cvtColor(img,cv2.COLOR_BGR2RGB), alpha=original_image_opacity) | |
| which_channel_dominates = np.zeros((img_arr.shape[0],img_arr.shape[1])) | |
| red_greater_green = np.greater(red_channel,green_channel) | |
| red_greater_blue = np.greater(red_channel,blue_channel) | |
| green_greater_blue = np.greater(green_channel,blue_channel) | |
| #Red is greatest if red is greater than green and blue | |
| which_channel_dominates[(red_greater_green & red_greater_blue)] = 1 | |
| which_channel_dominates[green_greater_blue & (~red_greater_green)] = 2 | |
| which_channel_dominates[~green_greater_blue & (~red_greater_blue)] = 3 | |
| which_channel_dominates[(red_channel == green_channel) & (red_channel == blue_channel)] = 6 | |
| which_channel_dominates[(red_channel == 255) & (blue_channel == 255) & (green_channel == 255)] = 4 | |
| which_channel_dominates[(red_channel == 0) & (blue_channel == 0) & (green_channel == 0)] = 5 | |
| print(np.unique(which_channel_dominates)) | |
| #Map the color code to the image | |
| plot = plt.imshow(which_channel_dominates, cmap=cmap, alpha=channel_opacity) | |
| #Customize the ticks of the colorbar | |
| plt.colorbar(plot, orientation='vertical', | |
| # ticks=[1,2,3,4,5,6], | |
| ticks = [], | |
| label='Dominant Color Channel' | |
| ) | |
| text = "Which channel dominates in the image below?\nWhite : R=G=B=255, Black : R=G=B=0\nGray : 0 < R=G=B< 255" | |
| plt.title(text) |