prob1b_single_frame.py

from matplotlib import pyplot as plt
import numpy as np
import cv2

img = cv2.imread("prob1_dataset/0000000000.png")
# Split blue, green and red channels of the image
# b, g, r = cv2.split(img)

def compute_histogram(image, bins=256):
    # Array with size of bins, set to zeros
    histogram = np.zeros(bins)
    # Loop through pixels and sum up counts of pixels
    for pixel in image:
        histogram[pixel] += 1
    # Return our final result
    return histogram

def cumsum(values):
    result = [values[0]]
    for i in values[1:]:
        result.append(result[-1] + i)
    return result

def equalize(entries):
    numerator = (entries - np.min(entries))*255
    denominator = np.max(entries) - np.min(entries)
    # Re-normalize the cdf
    result = numerator/denominator
    # Convert float to int
    result = result.astype('uint8')
    return result

def histogram_equalization(image):
    # Convert image into a numpy array
    image_array = np.asarray(image)
    # Convert array to into 1D array
    flatten_image = image_array.flatten()
    # Compute histogram
    computed_histogram_for_input = compute_histogram(flatten_image)
    # Compute cumulative sum 
    cumulative_sum = cumsum(computed_histogram_for_input)
    # Perform equalization over cumulative sum
    cumulative_sum_normalised = equalize(cumulative_sum)
    # Get the value from cumulative sum normalised for every index in flatten_image, and set that as computed_histogram_for_output
    computed_histogram_for_output = cumulative_sum_normalised[flatten_image]
    # Convert array back to original image shape
    final_image = np.reshape(computed_histogram_for_output,image.shape)
    return flatten_image, cumulative_sum, final_image, computed_histogram_for_output, cumulative_sum_normalised

def perform_equalization_and_merge_channels(img):
    # Split blue, green and red channels of the image
    b, g, r = cv2.split(img)
    flatten_image_b, cumulative_sum_b, result_b, histogram_equalized_b, cum_sum_norm_b = histogram_equalization(b)
    flatten_image_g, cumulative_sum_g, result_g, histogram_equalized_g, cum_sum_norm_g = histogram_equalization(g)
    flatten_image_r, cumulative_sum_r, result_r, histogram_equalized_r, cum_sum_norm_r = histogram_equalization(r)

    # Merge blue, green and red channels of the image
    merged_result = cv2.merge([result_b,result_g,result_r])
    return merged_result

block_img = np.zeros(img.shape,dtype=np.uint8)
print(img.shape)
# tile_size = 800
tile_size_x = 47
tile_size_y = 153
i=0
j=0
for j in range(j, img.shape[1], tile_size_y):
    i=0
    for i in range(i, img.shape[0], tile_size_x):
        tile = img[i:i+tile_size_x,j:j+tile_size_y,:]
        hist_tile = perform_equalization_and_merge_channels(tile)
        block_img[i:i+tile_size_x,j:j+tile_size_y,:] = hist_tile
        
cv2.imwrite('prob1_output/adaptive_histogram_equalization.png',block_img)
cv2.imshow('input',img)
cv2.imshow('result', block_img)
key = cv2.waitKey(0)
if key == 27:
    cv2.destroyAllWindows()