An architecture of an image processing system optimized for neural network
AN ARCHITECTURE OF AN IMAGE PROCESSING SYSTEM OPTIMIZED FOR NEURAL NETWORK. Sunao T a k a t o r i . R y o h e i Kumagai. Makoto Yamamoto. EZEL Inc. Re...
AN ARCHITECTURE OF AN IMAGE PROCESSING SYSTEM OPTIMIZED FOR NEURAL NETWORK. Sunao T a k a t o r i . R y o h e i Kumagai. Makoto Yamamoto. EZEL Inc. Recognition Technology Laboratory, Nihonseimei Yotsuya B i l d . ]6-2 Samon-cho,
Shinjuku-ku,
Tokyo 160,
Japan
An f e a t u r e of t h e b i o l o g i c a l visual information processing system is the optimized information compression or characteristics extraction in the i n f o r m a t i o n apprehension system as well as the information transmission system. Through this information compression, the cerebrum r e c e i v e s only highly abstracted and precious information processable by t h e c e r e b r u m o f r a t h e r low p r o c e s s speed. We h a v e a c c o m p l i s h e d an i m a g e p r o c e s s i n g s y s t e m o f an architecture s i m i l a r to t h e b i o l o g i c a l image processing system. This s y s t e m compresses immense v i s u a l information and e x t r a c t s c h a r a c t e r i s t i c s of an image then t r a n s m i t t s the compressed i n f o r m a t i o n or c h a r a c t e r i s t i c s to the central processing system or MPU, so that the MPU's process load is extremely decreased than the conventional image processing system. Consequently, a super high speed and wide range gener a l purpose image processing system is r e a l i z e d . These two problems of high speed and general purpose are considered u s u a l l y to be contrary to each other. For the new a r c h i t e c t u r e , we developed a couple of c h a r a c t e r i s t i c s e x t r a c t i n g ICs i n c l u d i n g the essence of EZEL's technology. T h e s e ICs h a v e h i g h p e r f o r m a n c e and flexibility because they are designed b a s e d on t h e conc e p t of b i o l o g i c a l image processing system. The ICs w i l l be important not only for improving conventional type image processing system but also for developing a v i s u a l system of the Neural Network.