In order to make an image processing system recognize an image in visual periphery, the following steps are to be carried out in sequence:
(1) Image pick-up,
(2) Image filtering and primary image preprocessing,
(3) More complex image post- processing to recognize a picked-up image.
Conventional image processing system in the market uses CCD/CMOS imager for image pickup and PC for preprocessing and post processing after converting analog signals picked-up by CCD/CMOS imager into digital signals without any significant data compression if any. This forceful calculation mechanism would inevitably increase the time and power consumption for processing
On the other hand, our innovative image processing system which utilizes IVS, performs both image pickup and image filtering concurrently on a Silicon Retina chip This on-chip preprocessing is done by flexible analog parallel calculation on image pixel data, the compressed image data from a Silicon Retina are then fed into FPGA, which efficiently carries out primary image preprocessing such as detection of the edge of a static image , and concurrently extract the contour of a moving object and the direction of it’s movement. Because of this architecture, IVS is very useful to carry out image processing in real time with very low power consumption.
By connecting IVS to PC, that the last phase of image post processing for image recognition can be carried out without much calculation load on PC
Characteristics of IVS system
The advantages and potentials of IVS unit are maximized in a large scale network to which many number of IVS units are connected.
In case of conventional CCD network camera system, some schemes such as use of broad-band dedicated line or application of data compression technology, are necessary to convey huge amount of digital data picked up by many CCD camera to a central PC which execute image processing of each CCD camera in the predetermined order
On the other hand, in the novel IVS system, each IVS unit operate independently and measures only necessary image characteristics information to be sent to a control PC.
As a result, large scale intelligent autonomous image processing system which includes mutual interaction among IVS units can be rather easily established by use of IVS.
