Computer vision systems come in all different shapes and sizes, with new systems being developed on an ongoing basis to serve a specific application. A computer vision system is defined by how it is going to be used. Its functionality may be application dependent or part of a sub-system of various other functions. Also defining a computer vision system is whether it is unchanging or if some of its functions can be learned or changed once in operation.
Various Functions of Computer Vision Systems
Many computer vision systems, while diverse for their working environment, have many functions in common. Typical features include pre-processing, image acquisition and extracting of image features. For example, all computer vision systems require some form of image sensor to produce a digital image. While this function is typical in computer vision systems, the methods may vary, ranging from light-sensitive cameras and ultra-sonic cameras to radar and tomography devices. Some computer vision systems also have more advanced functions like segmentation, detection and high-level processing to process shape, texture, or motion.
More Robust Computer Vision Systems On The Horizon
As advances are made in technology, scientists are applying them to functions and applications in computer vision. For example, new developments in imaging and embedded computing technology has led scientists to incorporate portable imagers into computer vision systems that can be connected via wireless networks. The result is a computer vision system for video surveillance or any other function that calls for activity to be noticed, processed and interpreted.
Technological advances also led a group of European researchers to develop HERMES, a cognitive computer vision system that consists of video cameras and software designed to recognize and predict human behavior, and translate it into ordinary human language. This type of application can be used for accident prevention, intelligent surveillance and psychology efforts. Artificial intelligence researchers are also investigating the use of computer vision systems to better interpret the natural movements of people. Scientists at Purdue University are focusing on heat diffusion within a computer vision system to better achieve 3D object recognition for complex shapes or when the shape of an object takes on another form. The system would scan an object’s surface and use algorithms to forecast how heat would be diffused throughout it. Since heat diffusion patters are predictable, the shape of the object can be determined regardless of its orientation. For example, a computer vision system of this type would be able to recognize the shape of a closed fist as a human hand.
More robust computer vision systems are on the horizon as French researchers make headway with their study on autonomic computing. This would allow for a computer vision system designed for one application to be easily applied to new applications in a totally different genre. Autonomic computing is being looked at in building computer vision systems that are less costly to install, perform self-monitoring functions and provide enhanced reliability.