In this paper it is shown that by building on parallel topographic CNN preprocessing of image flows, efficient terrain exploration and visual navigation algorithms can be developed. The approach combines several chann...
详细信息
In this paper it is shown that by building on parallel topographic CNN preprocessing of image flows, efficient terrain exploration and visual navigation algorithms can be developed. The approach combines several channels of nonlinear spatio-temporal feature detectors within an analogic CNN algorithm and produces unique binary maps of salient feature locations. This preprocessing scheme is embedded into a multi-target tracking (MTT) framework where these features are statistically described and assigned to numbered tracks. The MTT output has two distinct roles. First, its feature descriptors drive a classifier based on the adaptive resonance theory (ART), which is also implemented on CNN architecture. Second, it provides an optical flow ("target displacement") estimate to the navigation system, which in turn calculates the flight control parameters (Yaw-Pitch-Roll). An upper level visual attention and selection mechanism uses both the feature descriptors and the optical flow estimates to automatically adjust the focus and scale (zoom) during navigation. The paper describes the architecture and the algorithmic frameworks and provides the first experimental results on aerial video-flows.
This paper shows that the performance of multi-target tracking (MTT) systems can be significantly increased with stored program adaptive cellular array sensors. The primary motivation of the present work is to define ...
详细信息
This paper shows that the performance of multi-target tracking (MTT) systems can be significantly increased with stored program adaptive cellular array sensors. The primary motivation of the present work is to define a topographic microprocessor architecture for MTT with embedded sensors capable of operating in a process real-time manner. In the ongoing experiments it is assumed that the input data flow is acquired by a single array sensor and the data is processed on an adaptive CNN-UM architecture consisting of both a cellular nonlinear network (CNN) and digital signal processing (DSP) microprocessors. The algorithms designed for this combined hardware platform use adaptive multi-channel CNN solutions for instantaneous position estimation and morphological characterization of all visible targets and the DSP environment for distance calculation, gating, data association, track maintenance and dynamic target motion prediction. A special feature of the architecture is that it allows interactive communication between the sensor and the digital environment. The configuration of functional modules for various real-time applications is discussed. Using real-life video-flows, successful tracking of several maneuvering targets is demonstrated within the proposed adaptive multi-channel framework.
Different CNN-UM architecture implementations, analog and emulated digital, were developed. The emulated digital architecture (CASTLE) is accurate but slower than the analog CNN-UMs. It is generally disadvantageous es...
详细信息
Different CNN-UM architecture implementations, analog and emulated digital, were developed. The emulated digital architecture (CASTLE) is accurate but slower than the analog CNN-UMs. It is generally disadvantageous especially if transient computing is critical. The operation speed of the emulated digital implementations, namely CASTLE, can be increased significantly using the pipeline technique. This solution is analyzed with respect to area, time, etc. These arithmetic cores were tested and simulated using a VIRTEX FPGA development system.
One of the most critical errors is the short circuit in the manufacturing of printed circuit boards. In this contribution we extend the already existing solution to more general cases where the errors can be detected ...
详细信息
Newly emerging problems require high speed decision making based on visual perception of the environment. A project was set up to construct an intelligent agent like self-contained device that is capable to act in rea...
详细信息
ISBN:
(纸本)0780365682
Newly emerging problems require high speed decision making based on visual perception of the environment. A project was set up to construct an intelligent agent like self-contained device that is capable to act in real-time and show collaborative behavior. Giving the hardware basis for decisions to be made a cellular nonlinear network CNN (Chua and Yang, 1988) chip implementation's optical input is used in combination with the cooperative devices' information that is received via binary ports and serial ports. The proposed design is a self-contained compact device that is prepared to operate stand-alone for up to 10 hours running on medium sized batteries while doing measurements, logging and collaborating with its environment via parallel port (for image transfer), RS-232 port (using modbus, profibus, PPP protocols) and binary I/O-s. Intelligent power module, optical isolation, watch dog capability is also considered.
In this paper, experimental results on Cellular neural Network Universal Machine (CNN-UM) chips will be presented. These analogic spatio-temporal visual microprocessors make it possible that one can use nonlinear wave...
详细信息
In this paper, experimental results on Cellular neural Network Universal Machine (CNN-UM) chips will be presented. These analogic spatio-temporal visual microprocessors make it possible that one can use nonlinear wave...
详细信息
ISBN:
(纸本)0780366859
In this paper, experimental results on Cellular neural Network Universal Machine (CNN-UM) chips will be presented. These analogic spatio-temporal visual microprocessors make it possible that one can use nonlinear waves as the basic kernels of algorithms solving filtering-reconstruction and/or detection-classification problems. Showing output results from series of experiments it will be demonstrated how trigger waves, the simplest nonlinear waves, can constructively be used in a number of important application areas.
We present an analogic CNN algorithm that estimates the time to an impending collision between an approaching object and the observer. Calculation is based on a context insensitive method, which is well known in neuro...
详细信息
We present an analogic CNN algorithm that estimates the time to an impending collision between an approaching object and the observer. Calculation is based on a context insensitive method, which is well known in neurobiology, using only two specific cues of the expanding two-dimensional image of the looming object.
A simplified version of the gradient descent method is introduced as a straightforward way to find optimal 3/spl times/3 CNN templates for the inversion of known point spread functions (PSF). In practical applications...
详细信息
A simplified version of the gradient descent method is introduced as a straightforward way to find optimal 3/spl times/3 CNN templates for the inversion of known point spread functions (PSF). In practical applications the determination of this inverse is necessary to fulfil deconvolution tasks. The proposed method is much faster than the previously applied algorithms (like genetic algorithm) and still, in almost all practically important cases, it is convergent. Moreover, unlike a closed form method, it leads to 3/spl times/3 templates instead of 5/spl times/5 or bigger ones. In several important practical cases the PSF, which can be caused by motion, out of focus or the aberration of the imaging system, can be computed from object positions and from the optical system's parameters. Iterative deconvolution algorithms, which are necessary for volume reconstruction from microscopic image sequences, require considerable computation time. Using CNN-UM chips for these deconvolution tasks, a much higher speed, even real time processing seems to be achievable.
The objective of this paper is to provide a framework for the implementation of programmable Opto-Electronic Analogic CNN (POAC) computers embedding CNN Universal Chips. Specifically, a new method for optical CNN impl...
详细信息
The objective of this paper is to provide a framework for the implementation of programmable Opto-Electronic Analogic CNN (POAC) computers embedding CNN Universal Chips. Specifically, a new method for optical CNN implementation is provided and some details are experimentally studied. The POAC architecture includes the integration of an optical processing system, such as a joint transform correlator, with the fast spatio-temporal processing capabilities of a CNN-UM chip. We have built and tested an optical sub-unit of this experimental optoelectronic architecture to examine their processing capabilities for complex target recognition tasks. Preliminary result of these measurements will also be presented. The main idea is to introduce stored programmability into optical computing.
暂无评论