Computer vision (CV) is, very simply put, a method to recognize and interpret images using cameras and computers. Computer vision technology is utilized in a number of fields and is made up of a number of specialized hardware and software applications. Some types of computer vision technology include high-resolution cameras, individually designed computer systems, and specialty sensors or filters for both the camera and the computer.
Charged coupled device (CCD) cameras typically provide the image output for computer vision technology. CCD cameras can be omnidirectional, pan-tilt-zoom, or straight vision. Cameras developed by Carnegie Mellon University known as CMUcams are a type of computer vision technology that combine a video camera with a micro-controller. This allows for on-board support of simple image processing. Robotics often utilizes stereo vision, combining two cameras calibrated to capture an accurately converged image.
The computers used for computer vision technology purposes require special parts like daughter boards, also known as daughter cards, and processor boards designed to accelerate the design process. Sensors such as very large scale integration (VLSI) and infrared (IR) sensors are included to facilitate various tasks, such as night vision. Thermal sensors handle heat recognition.
Frame grabbers are implemented to take an analog image sent to the computer from the CCD camera or other image-capturing device and convert it to a digital image in gray-scale or color. Two-dimensional (2D) or three-dimensional (3D) line scanners are included as well, assisting in blob detection, motion sensing, and edge detecting. In certain applications, such as harsh environments, specialty enclosures may be used to protect the hardware.
Robotics and the security and surveillance industry are two of the primary fields using computer vision technology. The medical industry and astronomers play a big role as well. CCD cameras or the like provide the base image for the computer to process as requested by the programmer. Images can be processed generally, providing simple edge detection in 2D, which allows for motion estimation, or in 3D, which then allows for shape extraction.
All of the varying styles and configurations of computer vision technology utilize algorithms developed specifically for CV purposes. These algorithms assist with such tasks as enhancing images and finding lines to match them with models. The use of algorithms keeps the amount of data to be processed down to a minimum by extracting only the information necessary for a dedicated task.
While computer vision is constantly evolving in tandem with technology, it already plays an important part in the fields mentioned herein and many others. Blob detection and face recognition are important in security applications. Robotics relies on computer vision technology to maneuver successfully unmanned or autonomous vehicles. The current applications of the technology may be just the beginning of things that can be done with this emerging field of computer vision.