Computer vision
Computer vision develops methods that let computers extract information from images and video, including objects, motion, depth, and scene structure.
Image formation: The process by which light from a scene produces a measurable image through an optical system or sensor. Vision algorithms must account for how scenes become pixel measurements.
Digital image: A discrete representation of a visual scene, typically arranged as a grid of pixels with numerical values. Most computer vision methods begin with this sampled representation.
Autonomous vehicle: A vehicle that senses its environment and performs some or all driving tasks without direct human control. On-road autonomy depends on recognizing lanes, traffic participants, and hazards.
Computer graphics: The creation and manipulation of visual images using computers. Graphics generates images from models, while vision infers scene information from images.
Image segmentation: The partitioning of an image into regions or pixels associated with objects, materials, or other meaningful categories. It turns a dense image into regions that can be analyzed separately.
Feature extraction: The conversion of raw data into measurements or representations designed to capture informative patterns. Visual features provide representations for matching, recognition, and measurement.
Medical imaging: The use of technologies such as MRI, CT, and ultrasound to create images of internal anatomy and function. Vision methods help locate, classify, and measure findings in clinical images.
Image processing: The transformation or analysis of images using computational operations. Image processing often changes or enhances images; vision seeks information about their contents.
Object detection: The task of locating and classifying instances of objects in images or video. It combines object recognition with estimates of where objects appear.
Machine learning: Methods that enable computer systems to improve performance by learning patterns from data. Many modern vision systems learn their visual rules from examples rather than hand-coded instructions.