{ C } Digital Image Processing From Ground Up™

Download codeblocks and install it with the setup wizard, then confirm the GNU GCC compiler is detected; if you already have a C IDE or compiler, you may use it.

Open codeblocks and create a new C console project, then build and run a hello world to test the setup for DSP applications.

Explore real-world applications of image processing across medicine, defense, consumer electronics, and security, and learn how digital images are represented as 2D pixel grids with spatial and intensity data.

Explore image color and resolution, from binary and grayscale to RGB, and learn how spatial, temporal, and bit resolutions—and dynamic range—define pixel data and image quality.

Explore popular image formats—JPEG, GIF, BMP, PNG, and TIFF—and compare image data types, including binary, grayscale, RGB, and floating point images used in specialized fields.

Decode the bitmap format by reading the 54-byte header, the 1024-byte color table, and pixel data; extract width, height, and bit depth from offsets 18, 22, and 28.

Open a bitmap, read header and pixel data, and copy it to bmp file in a C console project. Extract width, height, bit depth, and color table for 8-bit images.

Modularize image processing by creating reusable image reader and writer functions that handle headers, color tables, and pixel data, then test by copying images.

Convert an rgb bitmap image to grayscale by combining red, green, and blue channels with weights 0.3, 0.59, and 0.11, and learn to read and write bmp headers as well.

Explore low-, mid-, and high-level image processing operations, including noise reduction, edge extraction, regions extractions, sharpening, contrast enhancement, and segmentation, to analyze and interpret image content.

Apply a threshold to a grayscale image to produce a binary black-and-white image, turning pixels to 0 or 255 and exploring how different threshold values shape results.

Explore image topology, neighborhood, adjacency, and connectivity in digital image processing, including 4-adjacent and 8-adjacent relations. The core takeaway is that the neighborhood means the pixels surrounding a given pixel.

Explore how a machine vision system for facial recognition processes images from acquisition and preprocessing to segmentation, feature extraction, and classification using neural networks and probabilistic classifiers.

Perform pixel-by-pixel arithmetic on images, including addition, subtraction, multiplication, and division, with additive image offset, overflow handling via normalization or truncation for 8-bit data.

Increase image brightness by adding a brightness factor to each pixel, saturating at 255, using a C program that reads the image header and color table and writes the result.

Subtract a darkness factor from each 8-bit pixel value (0-255) to darken an image, saturating any negative results at zero. Define the darkness factor and minimum color, apply the per-pixel operation, and generate new darker images to compare results.

Learn how image histograms show pixel value frequencies, reveal contrast, and distinguish foreground from background in eight-bit grayscale images, and prepare for histogram equalization.

Apply histogram equalization to transform a poor histogram into a flat histogram, boosting contrast so dark pixels appear darker and light pixels appear lighter, using p1(a), area1, and H1(a).

Learn how to download, install, and use the lightweight plotting tool can you upload to plot signals generated in C code and test your algorithms.

Learn how to install and launch gnu plot, accept the license, and verify plotting signals using both the console and patch level versions.

Learn to plot signals with gnuplot from algorithm output files, including input signals and impulse responses, using lines and colors.

Learn how to plot multiple signals in the same window using the new plot program. Configure quadrants and origins to compare input, impulse response, and output signals.

Learn to compute an image histogram by using the image reader to access pixel data, process header and color data, and write the histogram to a text file for plotting.

Learn histogram equalization to enhance image contrast by computing the image histogram from input image data and generating an equalized output image.

Learn modular brightness correction for a bitmap image by implementing a function that processes input image data to an output buffer using image dimensions and a brightness value.

Learn how adaptive thresholding sets a local threshold per pixel using its N×N neighborhood, with mean, median, or floor((max-min)/2) plus C to overcome global thresholding limits.

Learn how geometric operations reposition pixels to alter image geometry, using rotate, flip, crop, and resize to correct distortions, apply effects, and support image registration.

Explore affine transformation as a mapping from input image coordinates to output coordinates using a matrix; apply translation, scaling, rotation, and sharing to preserve straight lines.

learn to rotate an image by implementing a simple input and output file workflow that supports right, left, or 180-degree rotations using a 2x2 pixel buffer.

Discover grey level transformations to enhance images through simple pixel value mapping. Understand linear transformations such as s = c r + b that adjust contrast and brightness.

Create a negative image by reading an image, computing 255 minus each pixel, and writing the result to a file, while learning to manage image headers and color tables.

Improve image clarity by applying spatial domain filtering to reduce noise, sharpen edges, and create pleasing images, while respecting the information present and the chosen connectivity and local neighborhood.

Apply linear spatial filtering by sliding a kernel over the image and weighting neighborhood pixels. Sum the weighted neighborhood values to form the filtered image, i.e., via convolution.

Create a console project to blur a bitmap image using a 3x3 kernel by reading the header and color table and writing the result to output.

Design and implement a sepia filter by converting an input image with preset RGB coefficients and writing the result to an output file, noting BGR order.

Explore neighborhood processing, where a reference pixel and its neighbors determine the output via linear or nonlinear operations using 2D convolution with a mask.

Explore convolution and correlation as fundamental weight-based image processing operations on pixel neighborhoods. See a step-by-step one-dimensional example that demonstrates shifting, corresponding multiplication, and summing products to produce a result.

Explore 2-D convolution by sliding a mask over an image, flipping it in both dimensions, and summing products to produce blur, sharpen, or edge detection; compare to correlation without flipping.

Develop a 2-D discrete convolution algorithm in C to perform image filtering and edge detection using configurable masks, input images, and an output buffer.

Develop a line detection algorithm using a 3x3 line detector mask to identify vertical, horizontal, and diagonal lines, applying a dedicated line detector function to a Lena image.

Explore operators in image processing and implement edge-detection masks, including Pruitt's operator, the sol bell mask, the Robinson mask, and the Laplacian mask, using 2x2 and 3x3 convolutions to detect vertical, horizontal, and directional edges.

Learn how to perform edge detection with the Prewitt mask using a 3x3 kernel, implementing vertical and horizontal edge detectors via convolution on an image, producing new edge maps.

Apply the Sobel mask to perform edge detection with vertical and horizontal 3x3 kernels using 2D convolution on an input image.

Explore edge detection with the Robinson operator, a compass detector, using 3x3 masks to detect edges in north, south, east, and west directions.

This lesson demonstrates edge detection with the Kirsch mask, applying eight compass direction masks through convolution to produce directional edge maps from a cameraman image.

Apply edge detection with the Laplacian mask to a grayscale image, test positive and negative kernels, and compare responses using discrete convolution.

Practice edge detection with the Roberts cross detector using a 2x2 mask on grayscale images, computing g_x and g_y to reveal edges.