Computer Vision

Custom Search

Computer Vision

Computer vision is the artificial sense of sight used by robots (and other man-made systems) made possible by computer, microelectronic, and optical technologies. The technological field pertaining to this is also referred to as 'computer vision.'

Computer vision has been glamorized in many science fiction movies that depict intelligent robots that can 'see'. Although this application to robotic eyes is real, other computer vision applications exist. Computer vision technology, in its broad sense, deals with obtaining useful information from visual images, and therefore covers so many aspects of industry, education, and society.

It is only in the 1970's that computers became powerful and accessible enough to allow many researchers, scientists, and engineers to do serious work on computer vision. All the early developments in computer vision technology came from various fields and entities, and therefore lacked a standard hardware and software structure. In a way this has an advantage - there now exists a wide array of computer vision solutions in the market.

Computer vision can be used in many tasks, and it is the combination of these and other tasks that make up a whole application. Tasks where computer vision can be effectively used include but are not limited to the following:

- Object Recognition, wherein a computer vision system is used to find and identify objects in an image;

- motion tracking, wherein the movement of an object or set of objects (such as vehicles and people) is traced;

- optical character recognition (OCR), wherein printed, written, or drawn characters or images on a surface are identified and retrieved in digital form, usually for encoding into a format that a computer can use for an application;

- scene reconstruction, wherein a 3-D model of a scene is constructed from a given set of images;

- image restoration, enhancement, or alteration, wherein an image is digitally captured and then subjected to digital processing steps that changes it into a better or different image.

Figure 1. Industrial robots are equipped with computer vision to make them capable of more tasks.

A good computer vision system is capable of the following functions:

1) image acquisition, wherein the digital image is captured using one or more image sensors such as light-sensitive cameras, tomography devices, radar, ultrasonic cameras, etc.;

2) image pre-processing, wherein the image is subjected to image preparation techniques such as resampling, noise reduction, contrast enhancement, etc.

3) feature extraction, wherein various features of the image are extracted such as edges, ridges, corners, etc.

4) segmentation, wherein a certain region of the image that contains objects of interest is segmented for further processing;

5) high-level image processing, wherein a typically small portion of the image is subjected to high-level processing operations such as classification of an identified object into one of different categories or estimation of the actual size of an identified object.

Computer vision applications include: process control (including the use of computer vision in the control of industrial robots and autonomous vehicles), event detection (visual surveillance and people counting), management of visual information (organization of digital images), modeling of objects and environment, and computer-human interaction (such as those in robots that show emotions).