What is Computer Vision?
The term computer vision is often used to describe the use of computers to extract information from images. It may be concerned with the problems of creating or identifying pictures for general-purpose image processing, or it may be more specialized for purposes such as medical imaging, where clarity is necessary.
Computer vision can also refer to the capacity of a human operator to extract information about a target area using various technology systems. It usually describes high-tech solutions that require training and expertise in computer science and engineering – whereas optical techniques are more focused on how humans themselves see.
Technology that mimics human visual perception has the potential to perform a range of tasks to improve efficiency and safety for humans. There are, broadly, two main approaches:
“Physical” imaging uses information obtained through devices such as cameras and LIDARs. “Image” based imaging uses information based on images from the visible spectrum or infrared spectrum acquired by devices like satellites, airplanes, etc.
In terms of image processing, computer vision is very often used to refer to image recognition or machine vision which involves using techniques from graphical perception (see computer vision – history). These techniques involve features extraction and pattern analysis from the images.
A computer vision system relies on the digital images of an environment. The images can be in color or grayscale and from various sources such as TV cameras, security cameras, webcams, and image file formats.
Any computer vision system involves a number of concepts:
There are primarily two types of approaches to deal with the issues involved in computer vision. One approach is to use artificial intelligence techniques and try to learn from examples (see supervised learning). Any such system needs elaborate mathematical models which specify what inputs are meaningful and how outputs should be determined. These systems are generally implemented by machine learning algorithms.