5 Best Ways to Compute and Plot the Histogram of a Region in an Image with OpenCV and Python

💡 Problem Formulation: This article addresses the problem of calculating and visualizing the distribution of pixel intensities (histogram) within a specified region of interest (ROI) in an image using OpenCV with Python. For instance, suppose you have a photograph and you wish to analyze the grayscale value distribution of a specific area. The desired output is both the computed histogram data and its corresponding plot for the selected ROI.

Method 1: Using OpenCV’s `calcHist` Function

This method involves utilizing OpenCV’s calcHist() function to calculate the histogram of a specified region in an image. The function takes an image array, channels to analyze, a mask for the ROI, bin counts, and pixel value range, returning a histogram array highlighting the frequency of each pixel intensity in the ROI.

Here’s an example:

import cv2
from matplotlib import pyplot as plt

# Load the image in grayscale
image = cv2.imread('image.jpg', 0)

# Define region of interest (ROI) coordinates
# For instance, the top-left corner of the image
roi = image[0:100, 0:100]

# Create a mask with the same dimensions as the ROI
mask = cv2.inRange(roi, 10, 255)

# Compute histogram for the ROI
# 256 bins, range from 0-255
histogram = cv2.calcHist([image], [0], mask, [256], [0, 256])

# Plot the histogram
plt.figure()
plt.title('Histogram for ROI')
plt.xlabel('Pixel intensity')
plt.ylabel('Frequency')
plt.plot(histogram)
plt.xlim([0, 256])
plt.show()

Output of this code is a plot showing the distribution of pixel intensities in the ROI.

The above code snippet shows how to create a histogram for a specific ROI using OpenCV’s calcHist() function and then plotting it using Matplotlib. The mask is created to confine the histogram calculation to the ROI. The plot’s X-axis represents the pixel intensities, and the Y-axis shows their respective frequency.

Method 2: Normalizing Histograms with OpenCV

Normalization of histograms can be particularly helpful when comparing the distribution of pixel intensities of regions with different lighting conditions. OpenCV’s normalize() function scales the histogram bins such that the sum of all bins equals a specific value (often set to 1 for probability distribution).

Here’s an example:

import cv2
from matplotlib import pyplot as plt

# Load the image in grayscale
image = cv2.imread('image.jpg', 0)
roi = image[0:100, 0:100]

# Compute histogram for the ROI without mask this time
histogram = cv2.calcHist([roi], [0], None, [256], [0, 256])

# Normalize the histogram
cv2.normalize(histogram, histogram, alpha=0, beta=1, norm_type=cv2.NORM_MINMAX)

# Plot the normalized histogram
plt.figure()
plt.title('Normalized Histogram for ROI')
plt.xlabel('Pixel intensity')
plt.ylabel('Probability')
plt.plot(histogram)
plt.xlim([0, 256])
plt.show()

Output of this code is a plot showing the normalized distribution of pixel intensities in the ROI.

The code snippet demonstrates how to normalize a histogram using OpenCV’s normalize() function for the specified ROI and then plot it. Normalizing the histogram allows for a probability distribution representation, easing the comparison between different ROIs or images.

Method 3: Using the Masking Features in Matplotlib for Histograms

When plotting a histogram, Matplotlib provides masking functionality that can be useful for focusing on a region of interest within the image. The mask effectively selects the region in the image array for histogram computation.

Here’s an example:

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

# Load the image in grayscale
image = cv2.imread('image.jpg', 0)

# Create a mask with the same shape as the image, initially false
mask = np.zeros(image.shape[:2], dtype="uint8")

# Define the ROI and set it to true inside the mask
cv2.rectangle(mask, (0, 0), (100, 100), 255, -1)

# Apply the mask to the image
roi = cv2.bitwise_and(image, image, mask=mask)

# Use masked array to compute histogram
histogram = np.histogram(roi[mask == 255], bins=256, range=(0, 256))[0]

# Plot the histogram
plt.figure()
plt.title('Histogram for Masked ROI')
plt.xlabel('Pixel intensity')
plt.ylabel('Frequency')
plt.plot(histogram)
plt.xlim([0, 256])
plt.show()

Output of this code is a plot showing the histogram for the masked area of the image.

This snippet employs a mask to select the ROI for which we want to compute the histogram. This approach uses NumPy for histogram calculation and Matplotlib for plotting.cv2.bitwise_and() is used to apply the mask to the image and the Matplotlib suite plots the distribution of pixel intensities.

Method 4: Using 3D Histograms for Color Images

When working with color images, analyzing pixel distributions in 3-dimensional color spaces such as HSV or RGB can reveal more about image characteristics. This can be achieved in OpenCV by specifying different channels and appropriate bin sizes for each dimension.

Here’s an example:

import cv2
import numpy as np
from mpl_toolkits.mplot3d import Axes3D
from matplotlib import pyplot as plt

# Load the image in color
image = cv2.imread('image.jpg')
roi = image[0:100, 0:100]

# Convert ROI to HSV color space
hsv_roi = cv2.cvtColor(roi, cv2.COLOR_BGR2HSV)

# Compute the histogram in the HSV color space
hist = cv2.calcHist([hsv_roi], [0, 1, 2], None, [8, 8, 8], [0, 180, 0, 256, 0, 256])

# Plot the 3D histogram
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
for x, plane in enumerate(hist):
    for y, row in enumerate(plane):
        for z, col in enumerate(row):
            if col > 0:
                ax.scatter(x, y, z, s=col, c=[(x/8, y/8, z/8)], alpha=0.5)
ax.set_xlabel('Hue')
ax.set_ylabel('Saturation')
ax.set_zlabel('Value')
plt.show()

Output of this code is a 3D scatter plot showing the color distribution of the ROI in HSV space.

This method focuses on calculating and plotting a 3D histogram for a color image, which is helpful to analyze color distribution within a region. Here we convert ROI to the HSV color space and compute a 3D histogram. Then, we plot the histogram using Matplotlib’s 3D plotting capabilities.

Bonus One-Liner Method 5: Using List Comprehension and Pillow

For a quick and simple solution, Python’s Pillow library can be used in conjunction with list comprehensions to compute the histogram of an image’s region.

Here’s an example:

from PIL import Image
import numpy as np
from matplotlib import pyplot as plt

# Load the image
image = Image.open('image.jpg')

# Convert image to grayscale and get pixel data
pixels = list(image.convert('L').getdata())

# Define ROI coordinates
x, y, w, h = 0, 0, 100, 100

# Calculate histogram using a list comprehension
roi_pixels = [pixels[i * image.width + j] for i in range(y, y+h) for j in range(x, x+w)]
histogram = np.histogram(roi_pixels, bins=256, range=(0, 256))[0]

# Plot the histogram
plt.figure()
plt.title('Histogram for ROI using Pillow')
plt.xlabel('Pixel intensity')
plt.ylabel('Frequency')
plt.plot(histogram)
plt.xlim([0, 256])
plt.show()

Output of this code is a histogram plot created for the defined ROI using the Pillow library and list comprehension.

This one-liner approach uses Python’s Pillow library to read and manipulate the image, extracting pixel data for a defined ROI. List comprehensions offer a convenient way to grab pixel intensities from the ROI, which are then passed to NumPy’s histogram() function and plotted using Matplotlib.

Summary/Discussion

Method 1: OpenCV’s calcHist. This method provides robust functionality for histogram computation, including handling masks for ROIs. The complexity of handling masks may be a minor downside for some users.
Method 2: Normalizing Histograms. Offers a means for analysis and comparison of different images or ROIs, by normalizing histogram values. It does assume a prior calculation of the histogram.
Method 3: Masking with Matplotlib. Integrates closely with NumPy and Matplotlib for histograms and is quite versatile. However, the need to create a binary mask could add overhead to the process.
Method 4: 3D Histograms for Color Images. Well-suited for color image analysis, providing depth of information in color distribution, but may be computationally intensive.
Bonus Method 5: Pillow with List Comprehension. This is a simple and quick approach that utilizes Pillow, but it may not be as efficient or scalable for larger images or more complex histogram analysis tasks.

Method 1: Using OpenCV’s calcHist Function

Method 2: Normalizing Histograms with OpenCV

Method 3: Using the Masking Features in Matplotlib for Histograms

Method 4: Using 3D Histograms for Color Images

Bonus One-Liner Method 5: Using List Comprehension and Pillow

Summary/Discussion

Method 1: Using OpenCV’s `calcHist` Function