Computer vision is a branch of artificial intelligence that focuses on providing functionalities to computers which are typical
to human vision. It is mainly concerned with theory of building artificial systems that obtain information from images. The image data can be obtained in many ways like as video sequences,different views from cameras. Typical functionalities that can be expected from computer vision
applications are like controlling processes, detection of events, modeling object information. Application areas of computer vision are very wide like robotics, mechanics, automation, medical imaging, satellite imaging, bio metrics etc.
Research in this field had been going on since a long time on wider pace. In the initial days there was no that much technological support for this field in terms of powerful processing machines because even to run a small application we need to give millions of instructions per second. So these sort of applications can not be run on normal personal systems, for that the concept of parallel processing was introduced to increase the speed and efficiency of the systems by running multiple processors at the same time. But in recent years performance at system level increased by using faster micro processors, large memories and wider buses which made computer vision affordable on a wide scale.
Most of the computer vision applications have one thing in common, that is they tend to be either targeted to humans or controlled by humans. Vision systems have become the central sensor in the following applications.
->Human computer interfaces, where the application takes human interaction as the input and processes it gives the observation.
->Augmented perception, its an area where we will take aid of artificial vision tools where normal human vision is not enough to observe the events. Typically its aim is to increase the perception capability of human.
->Automatic media interpretation, where the system understands and analyzes the digital content like audio and video. Applications like automatic video summarization, sound source separation will come under this slab.
->Video
surveillance and bio metrics, this is one of more prominent slab of computer vision and currently many applications are already under usage in this sub area.
Main aim of Human computer interaction is to control application through computer using human gestures rather than keyboard input or mouse clicks. Sample application is like scrolling the computer screen up and down using human eye. This application mainly involves in eye lid motion tracking. Another application is using human nose as the mouse. There can be applications which can be controlled by human voice also. One such application was developed by Philips research labs, where voice is used perceive the surrounding environment. Camera will be there to sense the surrounds and it will convert that information from image to voice.
In media interpretation, main area where computer vision is used to annotate and index large multimedia data. Summarizing the important scenes from a big video, preparing highlights can be done in media interpretation. Computer vision applications can also be used in media transfer over the Internet, where we need to compress the large sized data without loosing much information. different multimedia compression techniques have been developed for this purpose. Techniques like super resolution can be used to enhance the low quality videos and images.
Video surveillance is one of the most developed and widely used application of computer vision. In olden days surveillance meant for low resolution and black and white videos which contains series of events. Nowadays computer vision facilitates the integration of views from multiple camera from multiple angles into one single better view. such a view automatically detects the scenes with people, vehicles and other targets of interest.. It also classifies them into different categories such as people, cars etc and extract their path ways, recognizes the different parts of it.
This analysis relies on a list of previously specified behaviors or on statistical observations such as differentiation of normal and abnormal behaviors. The main goal is not to completely replace security personnel but to assist them in supervising wider areas and focusing their attention on events of interest. There are some places where critical issue of privacy must be addressed before society widely adopts these video surveillance systems.
There are some systems for checking the pedestrian moment on parking lots, one such system was developed at University of technology in Sydney. Their approach is that a suspicious or abnormal behavior corresponds to an erroneous and random walking trajectory. Today's security world is mostly dependent on computer vision applications like under side vehicle scanners, license plate recognition systems, driver assistance systems. Its the field which has potential for application in every part of human life.
Researchers who want to build high end surveillance systems, an enormous amount of equipment is available nowadays starting from high resolution camera to whole human body scanners. Even the web servers are coming with special functionalities to transfer better quality multimedia data over the network which are captured in real time. Wireless technology has also been introduced to transfer the camera captured data over the Internet.
Computer vision giving opportunities to so many companies in the form wide variety of applications. This is the only field of computer science which can help the man kind in each and every part of life. Current machine equipment and technology demands more and more research in this field.