University of Waterloo

Image Segmentation/Classification

Extracting information from a digital image often depends on first identifying desired objects or breaking down the image into homogenous regions (a process called ‘segmentation’) and then assigning these objects to particular classes (a process called ‘classification’). This is a fundamental part of computer vision, combining image processing and pattern recognition techniques. Homogeneous may refer to the color of the object or region, but it also may use other features such as texture and shape. The methodology can be used to identify tumours in medical images, crops in satellite imagery, cells in biological tissue, or human faces in standard digital images or video. Each segmentation/classification implementation has the same fundamental approach; however, specific objects and imagery often require dedicated techniques for improved success. In the VIP lab, a dedicated example of segmentation is our advanced work in decoupled active contours. A dedicated example of classification is the automated identification of sea ice in satellite SAR images.

Related people




Related demos

Action Recognition in Video

Decoupled Active Contours

Disparate Scene Registration

Image Denoising

3D Reconstruction of Underwater Scenes

Skin Cancer Detection

Statistical Textural Distinctiveness for Salient Region Detection in Natural Images

Enhanced Decoupled Active Contour Using Structural and Textural Variation Energy Functionals

Computer Vision for Autonomous Robots

Hybrid Structural and Texture Distinctiveness Vector Field Convolution for Region Segmentation

MAGIC System

Grid Seams: A fast superpixel algorithm for real-time applications

VIP-Sal Dataset

Related publications

Journal Articles

Conference Papers