Add files via upload

Author: bnsreenu

112_denoising_images_by_dct_averaging.py

@@ -0,0 +1,72 @@
+#!/usr/bin/env python
+__author__ = "Sreenivas Bhattiprolu"
+__license__ = "Feel free to copy, I appreciate if you acknowledge Python for Microscopists"
+
+# https://youtu.be/XbCK46n7U80
+
+"""
+This code demonstrates potential applications in image processing by manipulating DCT.
+In this example, we perform averaging of images in real and DCT space. 
+Since DCT is linear, an average of images in spatial domain would be identical to performing
+DCT on each image and averaging the DCT and then inverse DCT to bring back the original image.
+
+DCT is similar to DFT. DFT uses a set of complex exponential functions, 
+while the DCT uses only (real-valued) cosine functions.
+
+DCT is used in data compression applications such as JPEG.This is becasue DCT representation
+has more of the information concentrated in small number of coefficients. This is useful for
+compression algorithms because you can approximate original information in a relatively small set
+of DCT coefficients. 
+"""
+
+import matplotlib.pyplot as plt
+import cv2
+import numpy as np
+from scipy.fftpack import dct, idct
+
+from os import listdir
+
+image_dir = "C:/PfM/Python_files/New_stuff/noisy_images/25_sigma/"  # Path to image directory
+
+# Load images
+filenames = listdir(image_dir)
+filenames.sort()
+
+imgs = []
+for f in filenames:
+	imgs.append((cv2.imread(image_dir + f, 0)).astype(np.float32))
+
+
+height, width = imgs[0].shape
+
+# Apply the weighted average to images and corresponding DCT images, respectively. 
+avg_img = np.zeros([height, width], np.float32)
+dct_avg_img = np.zeros([height, width], np.float32)
+
+for i in range(len(imgs)):
+    avg_img = cv2.addWeighted(avg_img, i/(i+1.0), imgs[i], 1/(i+1.0), 0) #Original image average
+    dct_avg_img = cv2.addWeighted(dct_avg_img, i/(i+1.0), dct(imgs[i]), 1/(i+1.0), 0) #DCT average
+
+
+reverse_img = idct(dct_avg_img) #Convert averaged DCT back to real space. 
+
+
+plt.imsave("noisy_images/00-dct_averaged_img.jpg", reverse_img, cmap="gray")
+plt.imsave("noisy_images/00-averaged_img.jpg", avg_img, cmap="gray")
+
+
+fig = plt.figure(figsize=(10, 10))
+ax1 = fig.add_subplot(2,2,1)
+ax1.imshow(imgs[0], cmap='gray')
+ax1.title.set_text('Input Image 1')
+ax2 = fig.add_subplot(2,2,2)
+ax2.imshow(imgs[1], cmap='gray')
+ax2.title.set_text('Input Image 2')
+
+ax3 = fig.add_subplot(2,2,3)
+ax3.imshow(avg_img, cmap='gray')
+ax3.title.set_text('Average of Images')
+ax4 = fig.add_subplot(2,2,4)
+ax4.imshow(reverse_img, cmap='gray')
+ax4.title.set_text('Image from DCT average')
+plt.show()

113-what_is_histogram_equalization.py

@@ -0,0 +1,74 @@
+#!/usr/bin/env python
+__author__ = "Sreenivas Bhattiprolu"
+__license__ = "Feel free to copy, I appreciate if you acknowledge Python for Microscopists"
+
+# https://youtu.be/jWShMEhMZI4
+
+"""
+Histogram equalization: 
+Stretches histogram to include all ranges if the original histogram is confined
+only to a small region - low contrast images. 
+But, this type of stretching may not result in ideal results and gives 
+too bright and too dark regions in the image. This can be very bad for images
+with large intensity variations. 
+
+CLAHE: COntrast limited adaptive histogram equalization
+Regular histogram equalization uses global contrast of the image. This results in
+too bright and too dark regions as the histogram stretches and is not confined
+to specific region.
+
+Adaptive histogram equalization divides the image into small tiles and within 
+each tile the histogram is equalized. Tile size is typically 8x8. 
+If theimage contains noise, it gets amplified during this process. Therefore, 
+contrast limiting is applied to limit the contrast below a specific limit.
+Bilinear interpolation is performed between tile borders. 
+
+Below, let us perform both histogram equalization and CLAHE and compare the results. 
+The best way to work with color images is by converting them to luminance space,
+e.g. LAB, and enhancing lumincnace channel only and eventually combining all channels. 
+    
+"""
+
+import cv2
+from skimage import io
+from matplotlib import pyplot as plt
+
+img = cv2.imread("images/bio_low_contrast.JPG", 1)
+#img = cv2.imread('images/retina.jpg', 1)
+
+#Converting image to LAB Color so CLAHE can be applied to the luminance channel
+lab_img= cv2.cvtColor(img, cv2.COLOR_BGR2LAB)
+
+#Splitting the LAB image to L, A and B channels, respectively
+l, a, b = cv2.split(lab_img)
+
+#plt.hist(l.flat, bins=100, range=(0,255))
+###########Histogram Equlization#############
+#Apply histogram equalization to the L channel
+equ = cv2.equalizeHist(l)
+
+#plt.hist(equ.flat, bins=100, range=(0,255))
+#Combine the Hist. equalized L-channel back with A and B channels
+updated_lab_img1 = cv2.merge((equ,a,b))
+
+#Convert LAB image back to color (RGB)
+hist_eq_img = cv2.cvtColor(updated_lab_img1, cv2.COLOR_LAB2BGR)
+
+###########CLAHE#########################
+#Apply CLAHE to L channel
+clahe = cv2.createCLAHE(clipLimit=3.0, tileGridSize=(8,8))
+clahe_img = clahe.apply(l)
+#plt.hist(clahe_img.flat, bins=100, range=(0,255))
+
+#Combine the CLAHE enhanced L-channel back with A and B channels
+updated_lab_img2 = cv2.merge((clahe_img,a,b))
+
+#Convert LAB image back to color (RGB)
+CLAHE_img = cv2.cvtColor(updated_lab_img2, cv2.COLOR_LAB2BGR)
+
+
+cv2.imshow("Original image", img)
+cv2.imshow("Equalized image", hist_eq_img)
+cv2.imshow('CLAHE Image', CLAHE_img)
+cv2.waitKey(0)
+cv2.destroyAllWindows()