Image Processing

From the basics to deep learning applications

About This Series

Image processing is the technical field of manipulating and analyzing digital images mathematically. This series starts from the basic concepts of images and proceeds step by step through filtering, edge detection, morphological processing, and on to advanced image recognition using deep learning.

Image processing has become an indispensable technology across many industries, including medical diagnosis, autonomous driving, surveillance systems, and quality inspection in manufacturing.

Learn by Level

Introductory

Image representation and tone (in preparation)

  • Sampling and quantization
  • Resolution (ppi, dpi) and bit depth
  • Gamma correction and tone curves
  • Interpolation and resizing
  • Moiré, aliasing, and file formats

Basic

Filtering and transforms (in preparation)

  • Convolution, correlation, and border handling
  • Smoothing and sharpening
  • Gradients, edge detection, DoG and LoG
  • The 2D Fourier transform (computed by FFT)
  • Noise models and noise removal

Intermediate

Image analysis (in preparation)

  • Adaptive thresholding
  • Morphology (top-hat, skeletonization, distance transform)
  • Hough transform and texture (GLCM, LBP)
  • Superpixels and active contours
  • Feature points and camera models: see Computer Vision

Advanced

Advanced methods and deep learning (in preparation)

  • Wavelet multiresolution
  • Total-variation denoising and compressed sensing
  • Optimal transport
  • Transposed convolution, self-attention, and diffusion models
  • Object-detection metrics, depth maps and point clouds

Learning Path

Intro Image basics Basic Filtering Interm. Image analysis Advanced Advanced & DL Intro: sampling, resolution, bit depth, gamma, interpolation Basic: convolution, smoothing/sharpening, edges, 2D Fourier, noise Interm.: thresholding, morphology, Hough, texture Adv.: wavelets, total variation, optimal transport, diffusion
Figure 1. Learning path: progress step by step from Introductory through Basic and Intermediate to Advanced (the main chapters of each level are in preparation).

How the Concepts Relate

Digital image Preprocessing Filtering Freq. transform Features Segmentation Deep learning Recognition
Figure 2. Concept map of image processing: from a digital image through preprocessing, filtering, and transforms to feature extraction, segmentation, and recognition. (Arrows show example relationships, not a fixed processing order; deep learning can take pixels directly as input.)

Main Topics

Image Basics

Pixels, sampling and quantization, resolution, bit depth and gamma correction, color spaces, and image formats.

Filtering

Convolution, smoothing and sharpening, edge detection, the 2D Fourier transform (computed by FFT), and noise removal.

Feature Extraction

Texture (GLCM, LBP), the Hough transform, Fourier descriptors, and shape features from morphology. Feature points (SIFT, ORB) are covered in the Computer Vision series.

Advanced Methods and Deep Learning

Wavelets, total variation and compressed sensing, optimal transport, transposed convolution, self-attention and diffusion models, and object-detection metrics.

Related Content

Computer Vision

The field of "images → 3D and semantic understanding": feature detection (SIFT, ORB), camera models, 3D reconstruction, SLAM, and more.

CT Image Processing

Content specialized in the reconstruction and analysis of medical CT scan images.

Frequently Asked Questions

What is image processing?

Image processing is the technology of using a computer to apply various transformations, analyses, and enhancements to digital images. Its applications are wide-ranging, including noise removal, edge detection, feature extraction, and segmentation. This series also covers CT reconstruction, the inverse problem of estimating cross-sectional images from projection data.

How are these image processing notes organized?

They are organized into four levels — introductory (intro), basic, intermediate, and advanced — plus a ct-imaging section specialized in CT scans. You can study step by step from the mathematical foundations through to implementation and applications (the main chapters of each level are in preparation).

What mathematical background is needed to study image processing?

The introductory level needs only high-school mathematics; higher levels increasingly use linear algebra (matrix and vector operations), calculus (partial derivatives, integral transforms), Fourier analysis, and probability and statistics. In CT imaging, knowledge of the Radon transform and of inverse problems and optimization (such as regularization) is also useful.

What is the difference between image processing and computer vision?

Image processing mainly takes an image as input and produces a better image or a more tractable representation through filtering, transforms, restoration, and segmentation. Computer vision aims to extract an understanding of the image's content, such as feature points, 3D structure, and the meaning of objects. The two form a continuum with substantial overlap, and topics like feature detection and camera models sit at the bridge between them.