Isolating and Modifying Color Channels

OpenCV loads images in the Blue-Green-Red (BGR) format by default. Direct manipulation of individual channels allows for specific color corrections or feature extraction.

import cv2
import numpy as np

def manipulate_channel_demo():
    # Load image from local path
    source_path = "../data/opencv2.png"
    img = cv2.imread(source_path)
    
    if img is None:
        raise FileNotFoundError(f"Image not found at {source_path}")

    # Display full original image
    cv2.imshow("Full Original", img)

    # Extract the Blue channel (index 0)
    # Shape indices: [height, width, channel]
    blue_channel = img[:, :, 0]
    
    # Display extracted channel (interpreted as grayscale)
    cv2.imshow("Blue Channel Only", blue_channel)

    # Suppress Blue channel influence by setting to zero
    modified_img = img.copy()
    modified_img[:, :, 0] = 0
    
    cv2.imshow("No Blue Component", modified_img)
    
    cv2.waitKey(0)
    cv2.destroyAllWindows()

if __name__ == "__main__":
    manipulate_channel_demo()

Enhancing Contrast with Hisotgram Equalization

Histogram equalization redistirbutes pixel intensity values to improve image contrast. This is critical for images with low dynamic range.

Grayscale Equalizatino

For single-channel inputs, the operation maps input cumulative distribution functions directly.

import cv2
from matplotlib import pyplot as plt

def grayscale_histogram_equalization():
    image_path = "../data/sunrise.jpg"
    
    # Read as grayscale matrix (0 means no color channels)
    gray_img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
    
    if gray_img is None:
        return

    # Apply histogram equalization algorithm
    eq_img = cv2.equalizeHist(gray_img)
    
    # Compare visual outputs
    cv2.imshow("Original Gray", gray_img)
    cv2.imshow("Equalized Gray", eq_img)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

    # Plot statistical distribution for analysis
    fig, ax = plt.subplots(2, 1, figsize=(10, 8))
    
    ax[0].hist(gray_img.ravel(), 256, [0, 256], label='Original')
    ax[0].set_title('Original Histogram Distribution')
    ax[0].legend()
    
    ax[1].hist(eq_img.ravel(), 256, [0, 256], color='r', label='Equalized')
    ax[1].set_title('Post-EQ Histogram Distribution')
    ax[1].legend()
    
    plt.tight_layout()
    plt.show()

Color Space Luminance Manipulation

In color images, modifying RGB channels directly alters hue and saturation artifacts. The recommended approach converts data to a luminance-aware space (like YUV) before processing.

import cv2

def color_luminance_enhancement():
    color_path = "../data/sunrise.jpg"
    
    original_color = cv2.imread(color_path)
    if original_color is None:
        return

    # Convert BGR to YUV
    yuv_image = cv2.cvtColor(original_color, cv2.COLOR_BGR2YUV)
    
    # Separate Y channel (Luminance)
    y_channel = yuv_image[:,:,0]
    
    # Equalize only the luminance component
    eq_y_channel = cv2.equalizeHist(y_channel)
    
    # Reconstruct image
    yuv_image[:,:,0] = eq_y_channel
    enhanced_bgr = cv2.cvtColor(yuv_image, cv2.COLOR_YUV2BGR)

    cv2.imshow("Enhanced Color Image", enhanced_bgr)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

Pixel-Level Segmentation via Thresholding

Thresholding transforms continuous intensity values into binary masks, separating foreground objects from background based on a cut-off value.

import cv2

def thresholding_operations():
    mask_path = "../data/lena.jpg"
    
    # Load as single channel
    src_img = cv2.imread(mask_path, cv2.IMREAD_GRAYSCALE)
    
    if src_img is None:
        return

    # Define threshold parameters
    threshold_val = 127
    max_val = 255

    # Standard Binary Thresholding
    # Pixels > threshold become max_val, others become 0
    _, binary_mask = cv2.threshold(src_img, threshold_val, max_val, cv2.THRESH_BINARY)
    
    # Inverse Binary Thresholding
    # Pixels > threshold become 0, others become max_val
    _, inverse_mask = cv2.threshold(src_img, threshold_val, max_val, cv2.THRESH_BINARY_INV)

    cv2.imshow("Binary Mask", binary_mask)
    cv2.imshow("Inverse Binary Mask", inverse_mask)
    
    cv2.waitKey(0)
    cv2.destroyAllWindows()