C. Koch and H. Li, Eds., Vision Chips: Implementing Vision Algorithms With Analog VLSI Circuits. Los Alamitos, CA: IEEE Computer Science Press, 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of on chip processing. Motivated by neural processes in biological retinas, a great majority of computational sensors implement local operations on a single light sensitive VLSI chip. Notably, there is a growing body of research in so called neuromorphic vision chips (for examples, see [10] and [12]) Typical examples of the operations implemented in these chips include spatial filtering and motion computation. Local operations produce preprocessed images ; therefore, a large quantity of data must be read out and further inspected before a decision for an appropriate action is ....

C. H. Koch and H. Li, Eds., Vision Chips: Implementing Vision Algorithms with Analog VLSI Circuits. New York: IEEE Computer Society, 1995.

....in category theory appear very pertinent to the formal philosophy of image representation and retrieval. Category theory provides a way to handle image objects consistent with parallel de 1 like opto electronics [5] holographs [13, 15] and artificial vision with neuromorphic vision chips [23]. 2 velopments in areas like signal processing, data warehousing, data mining and data fusion. The latter for the commercial environment seeks to integrate fully all the information for a particular business. Image data is now an important component needed to enhance the performance of many ....

C.Koch & H,Li, Vision Chips : Implementing vision Algorithms with Analog VLSI Circuits, IEEE Press (1994).

....including grey level and multiresolution processing, improved communication network, memory virtualisation . A. Global Architectural Choices Within the last decade, some remarkable analog circuit techniques have been developed for image pre processing with a few tens transistors per pixel [7]. But they are dedicated, whereas we look for versatility which requires programmability. CNNs (Cellular Non linear Networks) are an attempt to get analog programmability with one to two hundreds transistors per pixel [8] But CNN architectures do not scale well, because of precision and power ....

C.Koch and H.Li, ed., VISION CHIPS - Implementing Vision Algorithms with Analog VLSI Circuits, IEEE Comp. Soc. Press, 1995.

....VLSI. This mapping is possible because the physics of sub threshold operation and biological signal processing are identical. Using this approach, many VLSI chips which mimic the analog and parallel characteristics of visual, auditory and cortical neural circuits have been designed [Mead, 1989; Koch, 1995]. However, the effects of attempting to model biology accurately, coupled with the slow dynamics of sub threshold MOS circuits, have prevented early neuromorphic systems from capitalizing on the main benefit of silicon which is its speed. Only recently have researchers started to combine strong ....

....derivatives, multiplications and divisions. Hence, it is highly susceptible to noise and is unrealizable in compact hardware. The second method of image motion estimation is based completely on biological models, and it has been more successfully implemented than the computationally based model [Koch, 1995]. The biological models, which use primarily delayed local correlation, are excellent for direction detection, but they do not report speed explicitly, require large circuits and are limited to narrow spatiotemporal bandwidths. Therefore, to develop a general purpose image motion detection ....

Koch C. and H. Li (Eds.), Vision Chips: Implementing Vision Algorithms with Analog VLSI Circuits, IEEE Computer Press, 1995.

....consumption. Taking into consideration these requirements we developed our new hardware architecture. Current developments of optical flow sensor systems are focused on two different approaches: First, biologically motivated sensors with processing and sensor elements integrated on a single chip [5], where only prototypes with a few pixels (max. 26 x 26) have been fabricated; second, computer systems augmented with special circuits like correlation processors (e.g. 4] which are quite complex and hence, not suited for mobile vehicles. Based on an overview over optical flow algorithms ....

C. Koch, and H. Li. VISION CHIPS - Implementing Vision Algorithms with Analog VLSI Circuits. IEEE Computer Society Press, 1995.

No context found.

C. Koch and H. Li, Eds., Vision Chips: Implementing Vision Algorithms With Analog VLSI Circuits. Los Alamitos, CA: IEEE Computer Science Press, 1994.

....combined within one stage, so that sensors estimating the optical flow throughout a scene in real time can be conveniently integrated on a single chip. Such smart vision chips could be used in stand alone applications, such as tracking cars, or as front ends in conventional machine vision systems [1, 2]. Because image irradiance is a continuous function of time, asynchronous circuit implementations are preferable to clocked implementations. The latter introduce temporal aliasing artifacts that can significantly compromise time sensitive computations, such as those associated with optical flow. ....

C. Koch and H. Li, Eds., Vision Chips: Implementing Vision Algorithms with Analog VLSI Circuits. Los Alamitos: IEEE Computer Science Press, 1994.

....Parallel hardware implementations are particularly attractive if image acquisition and motion computation can be integrated on a single silicon chip. Such smart vision chips could be used in stand alone applications, such as tracking cars, or as front ends in conventional machine vision systems [1, 2]. Because image irradiance is a continuous function of time, asynchronous circuit implementations are preferable to clocked implementations. The latter introduce temporal aliasing artifacts that can significantly compromise time sensitive computations, such as those associated with optical flow. ....

....can be used in massively parallel architectures to evaluate integrative features of the optical flow. Given the robust, very compact, and low power nature of such smart vision sensors, they represent a true alternative for certain applications to more conventional digital image architectures [1]. More specifically, we have emphasized the following points: photoreceptor sizing is important because of the tradeoffs between spatial filtering and the reduction of noise. The limits of operation of motion sensors are determined by the maximum and minimum values of parameters in the system ....

C. Koch and H. Li, Eds., Vision Chips: Implementing Vision Algorithms with Analog VLSI Circuits. Los Alamitos: IEEE Computer Science Press, 1994.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC