Image Histograms in OpenCV

Image histograms are graphical representations of the pixel intensity distribution in an image. They are essential tools in image processing for understanding and manipulating image characteristics.

What is a Histogram?

A histogram is a graph showing the number of pixels in an image at each different intensity value. For an 8-bit grayscale image, there are 256 different possible intensities, and so the histogram will graphically display 256 numbers showing the distribution of pixels amongst those grayscale values.

Calculating Histograms

OpenCV provides several methods to calculate histograms:

1. Using calcHist()

import cv2
import numpy as np

# Read image
img = cv2.imread('image.jpg', 0)  # Read as grayscale

# Calculate histogram
hist = cv2.calcHist([img], [0], None, [256], [0, 256])

Parameters of calcHist():

• images: Source image(s) in square brackets
• channels: Index of channel (grayscale: [0], color: [0], [1], or [2])
• mask: Mask image for calculating histogram of specific regions
• histSize: Number of bins (usually [256])
• ranges: Range of pixel values [0, 256]

2. Using NumPy

hist, bins = np.histogram(img.ravel(), 256, [0, 256])

Color Histograms

For color images, we can calculate histograms for each color channel:

# Read color image
img = cv2.imread('image.jpg')

# Calculate histograms for each channel
color = ('b', 'g', 'r')
for i, col in enumerate(color):
    hist = cv2.calcHist([img], [i], None, [256], [0, 256])

Histogram Visualization

We can visualize histograms using Matplotlib:

import matplotlib.pyplot as plt

plt.hist(img.ravel(), 256, [0, 256])
plt.show()

For color histograms:

for i, col in enumerate(color):
    hist = cv2.calcHist([img], [i], None, [256], [0, 256])
    plt.plot(hist, color=col)
plt.xlim([0, 256])
plt.show()

Histogram Equalization

Histogram equalization is a method to improve image contrast by effectively spreading out the most frequent intensity values:

# Equalize histogram
equ = cv2.equalizeHist(img)

Adaptive Histogram Equalization (CLAHE)

CLAHE (Contrast Limited Adaptive Histogram Equalization) operates on small regions in the image, providing better results than standard histogram equalization:

# Create CLAHE object
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8,8))

# Apply CLAHE
cl1 = clahe.apply(img)

2D Histograms

2D histograms can be used to show the relationship between two channels:

# Convert to HSV
hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)

# Calculate 2D histogram
hist = cv2.calcHist([hsv], [0, 1], None, [180, 256], [0, 180, 0, 256])

Histogram Comparison

OpenCV provides several methods to compare histograms:

# Calculate histograms
hist1 = cv2.calcHist([img1], [0], None, [256], [0, 256])
hist2 = cv2.calcHist([img2], [0], None, [256], [0, 256])

# Normalize histograms
hist1 = cv2.normalize(hist1, hist1).flatten()
hist2 = cv2.normalize(hist2, hist2).flatten()

# Compare using different methods
d1 = cv2.compareHist(hist1, hist2, cv2.HISTCMP_CORREL)
d2 = cv2.compareHist(hist1, hist2, cv2.HISTCMP_CHISQR)
d3 = cv2.compareHist(hist1, hist2, cv2.HISTCMP_INTERSECT)
d4 = cv2.compareHist(hist1, hist2, cv2.HISTCMP_BHATTACHARYYA)

Histogram Backprojection

Histogram backprojection is a way of finding objects in an image using their histogram:

# Convert to HSV
hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
target_hsv = cv2.cvtColor(target, cv2.COLOR_BGR2HSV)

# Calculate histogram of target
roihist = cv2.calcHist([target_hsv], [0, 1], None, [180, 256], [0, 180, 0, 256])

# Normalize histogram
cv2.normalize(roihist, roihist, 0, 255, cv2.NORM_MINMAX)

# Calculate backprojection
dst = cv2.calcBackProject([hsv], [0, 1], roihist, [0, 180, 0, 256], 1)

Practical Applications

1. Image Enhancement

def enhance_image(image):
    # Split the image into color channels
    b, g, r = cv2.split(image)
    
    # Apply CLAHE to each channel
    clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8,8))
    b = clahe.apply(b)
    g = clahe.apply(g)
    r = clahe.apply(r)
    
    # Merge the channels back
    enhanced = cv2.merge([b, g, r])
    return enhanced

2. Object Detection using Histogram Backprojection

def detect_object(image, target):
    # Convert images to HSV
    hsv = cv2.cvtColor(image, cv2.COLOR_BGR2HSV)
    target_hsv = cv2.cvtColor(target, cv2.COLOR_BGR2HSV)
    
    # Calculate target histogram
    target_hist = cv2.calcHist([target_hsv], [0, 1], None, [180, 256], [0, 180, 0, 256])
    cv2.normalize(target_hist, target_hist, 0, 255, cv2.NORM_MINMAX)
    
    # Calculate back projection
    back_proj = cv2.calcBackProject([hsv], [0, 1], target_hist, [0, 180, 0, 256], 1)
    
    # Apply threshold
    _, thresh = cv2.threshold(back_proj, 50, 255, cv2.THRESH_BINARY)
    
    return thresh

3. Image Matching using Histogram Comparison

def match_images(image1, image2):
    # Calculate histograms
    hist1 = cv2.calcHist([image1], [0, 1, 2], None, [8, 8, 8], [0, 256, 0, 256, 0, 256])
    hist2 = cv2.calcHist([image2], [0, 1, 2], None, [8, 8, 8], [0, 256, 0, 256, 0, 256])
    
    # Normalize histograms
    cv2.normalize(hist1, hist1, 0, 1, cv2.NORM_MINMAX)
    cv2.normalize(hist2, hist2, 0, 1, cv2.NORM_MINMAX)
    
    # Compare histograms
    similarity = cv2.compareHist(hist1, hist2, cv2.HISTCMP_CORREL)
    return similarity

4. Automatic Thresholding using Histogram Analysis

def auto_threshold(image):
    # Calculate histogram
    hist = cv2.calcHist([image], [0], None, [256], [0, 256])
    
    # Find peak values
    peak1 = np.argmax(hist[:128])
    peak2 = np.argmax(hist[128:]) + 128
    
    # Calculate threshold as midpoint
    threshold = (peak1 + peak2) // 2
    
    # Apply threshold
    _, binary = cv2.threshold(image, threshold, 255, cv2.THRESH_BINARY)
    return binary

Conclusion

Histograms are powerful tools in image processing that can be used for:

• Understanding image characteristics
• Improving image contrast
• Object detection and tracking
• Image matching and comparison
• Automatic thresholding and segmentation

In the next tutorial, we'll explore video basics in OpenCV, including how to capture, process, and save video streams.