Real-time Motion Detection Research Articles

2P2-B13 高速繰り返し動作における異常現象の実時間検出システムの開発(ロボットビジョン)

2P2-B13 高速繰り返し動作における異常現象の実時間検出システムの開発(ロボットビジョン)

Real-time motion detection with a mixed analogue–digital neuromorphic vision system

Abstract. We fabricated a novel vision system consisting of a neuromorphic silicon retina and FPGA circuits. The silicon retina executes two classes of fundamental image pre-processing: a Laplacian–Gaussian-like spatial filtering and a subtraction of consecutive image frames. The output images of the silicon retina are fed to FPGA circuits, with which image cues, i.e., contours, motion, direction of motion and centre of moving objects, are extracted within a single frame sampling time of the silicon retina. The system has compact hardware and low power consumption and therefore is suitable for controlling autonomous robots.

Low-Cost Real-Time 2-D Motion Detection Based on Reconfigurable Computing

This paper presents a method for real-time hand motion detection in two-dimensional (2-D) space. A new input device for kinetically challenged persons that uses this method is presented. The device consists of a solid-state accelerometer that senses 2-D motion, a microcontroller that samples the data in real time, and an embedded field-programmable gate array (FPGA) device that distinguishes the types of motion from programmable motion vocabularies. The system has a quadratic capability O(n <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ) in detecting motions, while the hardware used has a linear-complexity O(n). The motion-detection computational model is presented, along with experimental results. The system adaptation to individual requirements and the cost versus quality tradeoff can be addressed through reconfiguration

Three-Dimensional Tracking Method of Tissue Motion with Biplane Images

For real-time three-dimensional tissue motion detection using biplane ultrasound images, we investigated the efficiency of a correlation method in which each region of interest (ROI) is divided into sub-regions so small that the deformation in each subregion can be ignored when estimating its motion in and perpendicular to the image planes. The allowed range of displacement perpendicular to the image planes in sensitively estimating tissue motion by cross-correlation was analyzed. In vitro results with a phantom revealed that the velocity component in the image plane could be detected when the perpendicular displacement between two consecutive frames was as small as 0.4 mm. Moreover it was also revealed that the subregion size must be larger than the speckle size. We also performed in vivo experiments on human liver and found that the tissue motion induced by systole was detectable by our cross-correlation method. The obtained results demonstrated the potential usefulness of the proposed method in medical applications, such as an image guide for minimally invasive therapy.

A Simple-Architecture Motion-Detection Analog VLSI Based on Quasi-Two-Dimensional Hardware Algorithm

Analog VLSI circuits have been developed aiming at real-time motion detection in a very simple hardware configuration. The chip handles motion pictures captured in a two-dimensional photo sensor array not by a pixel-wise processing but by projecting the pixel data onto principal axes. Since the data processing is carried out only for projected data sets, all computing circuits are placed at the periphery of the photo sensor array, thus achieving a very area-efficient implementation. By introducing the concept of time derivative to the projection data, the background scenery information is erased and the motion of objects can be detected robustly. In order to confirm the concept, test chips were fabricated in a 0.6-μm double-poly triple-metal CMOS technology and real time motion detection has been experimentally demonstrated.

Real-time motion detection of functional MRI data

Real-time motion detection of functional MRI data

A CMOS focal-plane motion sensor with BJT-based retinal smoothing network and modified correlation-based algorithm

This work presents and implements a CMOS real-time focal-plane motion sensor intended to detect the global motion, using the bipolar junction transistor (BJT)-based retinal smoothing network and the modified correlation-based algorithm. In the proposed design, the BJT-based retinal photoreceptor and smoothing network are adopted to acquire images and enhance the contrast of an image while the modified correlation-based algorithm is used in signal processing to determine the velocity and direction of the incident image. The deviations of the calculated velocity and direction for different image patterns are greatly reduced by averaging the correlated output over 16 frame-sampling periods. The proposed motion sensor includes a 32/spl times/32 pixel array with a pixel size of 100/spl times/100 /spl mu/m/sup 2/. The fill factor is 11.6% and the total chip area is 4200/spl times/4000 /spl mu/m/sup 2/. The DC power consumption is 120 mW at 5 V in the dark. Experimental results have successfully confirmed that the proposed motion sensor can work with different incident images and detect a velocity between 1 pixel/s and 140,000 pixels/s via controlling the frame-sampling period. The minimum detectable displacement in a frame-sampling period is 5 /spl mu/m. Consequently, the proposed high-performance new motion sensor can be applied to many real-time motion detection systems.

Fast motion detection in compressed domain for video surveillance

A fast algorithm to detect motion in the compressed domain for video surveillance is presented. The algorithm partially decompresses the video bit stream and performs motion detection from its quantised discrete cosine transform coefficients. The very low computational cost makes this algorithm very useful when real-time motion detection has to be performed simultaneously in several video bit streams.

Biologically Inspired Visual Motion Detection in VLSI

Visual motion detection is a fundament component of vision, and plays a vital role in scene analysis and understanding for behaving organisms. In computer vision, motion detection requires considerable resources to obtain real-time results. Very Large Scale Integration (VLSI) technology offers a convenient substrate upon which both photosensitive elements and motion extracting circuits can be implemented, thus allowing real-time motion detection. This paper presents two approaches for implementing real-time visual motion detection in VLSI. The two approaches mimic the two primary methods found in biological organisms. Insect motion detection employs local correlation and is implemented very close to the photoreceptors. In contrast, primate motion detection is performed in cortex, using spatiotemporally oriented neural filters. The analysis, construction and results of the hardware models of insect and primate visual motion detection are presented.

Modeling and characterization of logarithmic complementary metal–oxide–semiconductor active pixel sensors

We present a detailed analysis of logarithmic active pixel sensors (log-APS) to be used in complementary metal–oxide–semiconductor (CMOS) imagers for real-time on-chip motion detection. Based on an equivalent circuit model for CMOS-compatible photodiodes, an HSPICE simulation has been used to characterize different configurations of these sensors under various conditions of light intensities and switching speeds. These investigations are supported by the experimental results obtained from the chip fabricated with standard 0.5 μm CMOS technology. It is concluded that more robust on-chip motion detection in CMOS imagers can be achieved with careful design of its log-APS photocircuits that considers the issues discussed here.

Robust real-time periodic motion detection, analysis, and applications

We describe new techniques to detect and analyze periodic motion as seen from both a static and a moving camera. By tracking objects of interest, we compute an object's self-similarity as it evolves in time. For periodic motion, the self-similarity measure is also periodic and we apply time-frequency analysis to detect and characterize the periodic motion. The periodicity is also analyzed robustly using the 2D lattice structures inherent in similarity matrices. A real-time system has been implemented to track and classify objects using periodicity. Examples of object classification (people, running dogs, vehicles), person counting, and nonstationary periodicity are provided.

Parallel nonlinear optoelectronic image processing for real-time motion detection

An analog nonlinear optoelectronic image-processing system for detection of moving objects in an input image stream is proposed and analyzed. Its operation is based on the effect of self-suppression of a stationary-state image background, with enhancement of the contrast of a moving object. Analytical results obtained in the linear approximation are verified by direct numerical simulations. © 1999 Society of Photo-Optical Instrumentation Engineers. (S0091-3286(99)02203-5)

Spatio-temporal dynamics of networks of heart cells in culture

Spontaneous activity of cultured rat ventricular heart cells is studied as a function of cell density and ambient temperature. Three complementary methods are used: calcium fluorescence microscopy (monitoring intracellular free Ca 2+ concentrations), real-time motion detection and electrophysiology. Statistical features of beating activity in single cells, small groups of cells and large monolayered networks are characterized and compared. At relatively high temperatures, a gradual transition from stochastic activity at the level of a single cell to regular synchronized beating in large, dense networks is found. Lowering the temperature results in a gradual loss of synchronization and the emergence of spatio-temporal dynamics, expressed by bursts of activity and contractions on many length and time scales.

Motion tracking with an active camera

This paper describes a method for real-time motion detection using an active camera mounted on a pan/tilt platform. Image mapping is used to align images of different viewpoints so that static camera motion detection can be applied. In the presence of camera position noise, the image mapping is inexact and compensation techniques fail. The use of morphological filtering of motion images is explored to desensitize the detection algorithm to inaccuracies in background compensation, Two motion detection techniques are examined, and experiments to verify the methods are presented. The system successfully extracts moving edges from dynamic images even when the pan/tilt angles between successive frames are as large as 3. >

Headreader: Realtime motion detection of human head from image sequence

AbstractHuman motion plays a very important role in daily communication. If computers can directly read the meaning of human motion, man‐machine interaction will be improved. This paper presents a three‐dimensional head motion detection system called “headreader.” A simple but very fast algorithm is used which exploits the contrast of hair and face to recognize face orientation. The system extracts the head and face area and then estimates the head motion parameters from the change in position of each area centroids. The head motion is computed at nearly seven frames per second on an SUN4 workstation.