Basic image manipulation with arrays in NumPy

Introduction

We use arrays to change images because images are just grids of numbers. Changing these numbers changes the picture.

You want to make a photo brighter or darker.

You want to flip or rotate a picture.

You want to crop a part of an image.

You want to change colors or add filters to a photo.

Syntax

NumPy

import numpy as np

# Load or create an image as a 2D (grayscale) or 3D (color) numpy array
image = np.array([...])  # Example image array

# Example operations:
# 1. Brightness change
brighter_image = np.clip(image + 50, 0, 255)  # Add 50 to all pixels and clip to valid range

# 2. Flip image horizontally
flipped_image = np.fliplr(image)

# 3. Crop image
cropped_image = image[10:100, 20:150]

# 4. Convert to grayscale (if color image)
gray_image = np.mean(image, axis=2).astype(np.uint8)

Images are stored as arrays where each number represents a pixel's brightness or color.

Use numpy functions to easily manipulate these arrays.

Examples

This creates a 5x5 black image (all pixels zero).

NumPy

import numpy as np

# Empty image (all zeros)
empty_image = np.zeros((5, 5), dtype=np.uint8)
print(empty_image)

An image with only one white pixel.

NumPy

import numpy as np

# Single pixel image
single_pixel_image = np.array([[255]], dtype=np.uint8)
print(single_pixel_image)

Flips the image left to right.

NumPy

import numpy as np

# Flip a small image horizontally
small_image = np.array([[10, 20, 30], [40, 50, 60]], dtype=np.uint8)
flipped = np.fliplr(small_image)
print(flipped)

Crops a part of the image; numpy handles out-of-bound indexes safely.

NumPy

import numpy as np

# Crop image edge case: cropping beyond image size
image = np.arange(100).reshape(10, 10)
cropped = image[8:15, 8:15]
print(cropped)

Sample Program

This program creates a small grayscale image as a numpy array. It then makes the image brighter, flips it horizontally, and crops the center part. Each step prints the image array so you can see the changes.

NumPy

import numpy as np

# Create a simple 5x5 grayscale image with values from 10 to 210
original_image = np.array([
    [10, 50, 90, 130, 170],
    [20, 60, 100, 140, 180],
    [30, 70, 110, 150, 190],
    [40, 80, 120, 160, 200],
    [50, 90, 130, 170, 210]
], dtype=np.uint8)

print("Original Image:\n", original_image)

# Increase brightness by 40 (clip max to 255)
brighter_image = np.clip(original_image + 40, 0, 255).astype(np.uint8)
print("\nBrighter Image:\n", brighter_image)

# Flip image horizontally
flipped_image = np.fliplr(original_image)
print("\nFlipped Image (Left-Right):\n", flipped_image)

# Crop the center 3x3 area
cropped_image = original_image[1:4, 1:4]
print("\nCropped Image (3x3 center):\n", cropped_image)

Important Notes

Brightness change operation runs in O(n*m) time where n and m are image dimensions.

Flipping and cropping also run in O(n*m) time but are very fast with numpy.

Common mistake: Not clipping values after brightness change can cause overflow and wrong colors.

Use cropping to focus on parts of an image; use flipping to mirror images easily.

Summary

Images are arrays of numbers representing pixels.

You can change images by adding, flipping, or slicing these arrays.

Numpy makes these operations simple and fast.