Adaptive invariant novelty filters

Szu, Harold; Messner, Richard A.

doi:10.1109/proc.1986.13494

Cited by 23 publications

(11 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Altogether there are millions of neurofibers that are closely packed together into a uniform output bundle toward the lateral geniculate nucleus. 24 For example, the input i'th pixel of facial boundary is r i = exp(u j ), which is equivalent to the output u j = log(r i ) achieving a massive parallel flow-through logarithmic scaling transform without explicit computation. In other words, when a face changes its size with a scale factor s at r i ≡ r i s = exp(u i ) implying u i = log(r i ) = log(r i s) = log(r i ) + log(s) ∼ = u j .…”

Section: Solving Variable Faces By Focusing At Thementioning

confidence: 99%

“…For example, a size of half s = 2 change gives log e (2) = ln(2) = 0.69 which is a small fraction over many pixels. This fanin polar exponential grid (PEG) architecture 24 facilitates the logarithmic computation of millions of rods' coordinates in a parallel flow through the PEG. This real-time "algo-tecture" is one of the wonders of the human visual system.…”

Section: Solving Variable Faces By Focusing At Thementioning

confidence: 99%

“…A preprocessing of viewing transformation and registration for each facial box is necessary before time-ordering reconstruction. This preprocessing is introduced to overcome the change of facial boxes by a maximum overlapping central region by a morphological image processing focusing the binarized center of each face box to achieve the scale invariant peripheral vision 24 in Sec. 3.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Re-establish the time-order across sensors of different modalities

Hsu

2011

Opt. Eng

View full text Add to dashboard Cite

Abstract. Modern cameras can cut passengers' faces into boxes in 0.04 s per frame in parallel without time stamps. Unfortunately, that creates random storage without the tracking capability, and one can no longer meet the 5 W's challenge-"who speaks what, where and when." We develop a time-order reconstruction methodology which sorts the boxes as follows. i. A morphological image preprocessing to overcome the facial changes is based on the peripheral invariance of a human visual system when focusing on a maximum overlapping central region. ii. Replacing the Wiener matched filter desired output with an averaged but blurred long exposure, one can select the best matched sharp short exposures called the anchor faces β's. iii. The time-order neighborhood chaining is done by an iterative self-affirmation logic that demands a mutually agreedupon minimum distance: whether or not the two nearest neighbors of β, namely face A and face C, also consider β to be their two nearest neighbors. The reconstruction procedure mathematically amounts to a product of two triple correlation functions sharing an intermediate state. We have thus demonstrated the time-order helps us associate a video submanifold with the acoustic manifold that solves the 5 W's challenge. C 2011 Society of Photo-Optical Instrumentation Engineers (SPIE).

show abstract

Section: Solving Variable Faces By Focusing At Thementioning

confidence: 99%

Section: Solving Variable Faces By Focusing At Thementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Re-establish the time-order across sensors of different modalities

Hsu

2011

Opt. Eng

View full text Add to dashboard Cite

show abstract

“…This fact allows for more real-time image and video processing to be accomplished due to the reduced dataset. Some machine vision application areas where this has been exploited include: optical flow [8], time-to-collision and depth from motion [9]; target tracking [10], [11]; pattern recognition [12]; image registration [13]; spacecraft docking [14]; front-end for space-variant image processing [15], [16].…”

Section: Introductionmentioning

confidence: 99%

Biologically motivated composite image sensor for deep-field target tracking

Melnyk

Messner

2007

SPIE Proceedings

View full text Add to dashboard Cite

The present work addresses the design of an image acquisition front-end for target detection and tracking within a wide range of distances. Inspired by raptor bird's vision, a novel design for a visual sensor is proposed. The sensor consists of two parts, each originating from the studies of biological vision systems of different species. The front end is comprised of a set of video cameras imitating a falconiform eye, in particular its optics and retina [1]. The back end is a software remapper that uses a popular in machine vision log-polar model of retino-cortical projection in primates [2], [3], [4]. The output of this sensor is a composite log-polar image incorporating both near and far visual fields into a single homogeneous image space. In such space it is easier to perform target detection and tracking for those applications that deal with targets moving along the camera axis. The target object preserves its shape and size being handled seamlessly between cameras regardless of distance to the composite sensor. The prototype of proposed composite sensor has been created and is used as a front-end in experimental mobile vehicle detection and tracking system. Its has been tested inside a driving simulator and results are presented.

show abstract

“…Several extensions were proposed by Shepanski (1988) which included the use of multiple layers and a clever encoding scheme to capture nonlinear mapping in only a few iterations. Several optical/electro-optical implementations have also been reported (Szu and Messner, 1986;Tekolste and Guest, 1987). To date, the two primary uses of the optimal linear assaciative memory are for pattern matching and novelty detection.…”

Section: Introductionmentioning

confidence: 99%

On reducing the influence of noise in a new model for optimal linear associative memory

Tuan

Li²,

Yau³

et al. 1991

[Proceedings] 1991 IEEE International Joint Conference on Neural Networks

View full text Add to dashboard Cite

We propose a new model for linear associative memory. In this work, a synaptic matrix composes of not only the stored input and output patterns, but also of injected attached patterns (constructed by designers) which are weighted periodic inverse-repeat pseudo random patterns.When the injected patterns are taken to be the stored input patterns, one obtains Kohonen's model. As such, Koh* nen's model is a special case of the model proposed here. Furthermore, when the real pattern is contaminated by (different kinds of) color noise, recalling the stored pattern is superior than that obtained from Kohonen's pseudo-inverse learning rule: our learning rule can reduce the color noise influence on the optimal linear associative memory.We believe that injecting attached patterns to neurons is a new vista in neural network theory. The learning rule proposed here is shown to be optimal in the least mean square sense. We illustrate the theoretical results with specific computer simulation.the Department of Mathematics, Harvard University for providing the initial graphical data. Prof Alan Yuille from the Department of Computer Science and John Yao from the Division of Applied Sciences, Harvard University, have improved greatly this manuscript with many helpful advises. The algorithms for this model were implemented by Dr Bainan Li in a SUN work station in the computer laboratories of the

show abstract

Adaptive invariant novelty filters

Cited by 23 publications

References 5 publications

Re-establish the time-order across sensors of different modalities

Re-establish the time-order across sensors of different modalities

Biologically motivated composite image sensor for deep-field target tracking

On reducing the influence of noise in a new model for optimal linear associative memory

Contact Info

Product

Resources

About