Fast Motion Detection Research Articles

Learning heterogeneous delays in a layer of spiking neurons for fast motion detection.

The precise timing of spikes emitted by neurons plays a crucial role in shaping the response of efferent biological neurons. This temporal dimension of neural activity holds significant importance in understanding information processing in neurobiology, especially for the performance of neuromorphic hardware, such as event-based cameras. Nonetheless, many artificial neural models disregard this critical temporal dimension of neural activity. In this study, we present a model designed to efficiently detect temporal spiking motifs using a layer of spiking neurons equipped with heterogeneous synaptic delays. Our model capitalizes on the diverse synaptic delays present on the dendritic tree, enabling specific arrangements of temporally precise synaptic inputs to synchronize upon reaching the basal dendritic tree. We formalize this process as a time-invariant logistic regression, which can be trained using labeled data. To demonstrate its practical efficacy, we apply the model to naturalistic videos transformed into event streams, simulating the output of the biological retina or event-based cameras. To evaluate the robustness of the model in detecting visual motion, we conduct experiments by selectively pruning weights and demonstrate that the model remains efficient even under significantly reduced workloads. In conclusion, by providing a comprehensive, event-driven computational building block, the incorporation of heterogeneous delays has the potential to greatly improve the performance of future spiking neural network algorithms, particularly in the context of neuromorphic chips.

Fabrication of a dual-focus artificial compound eye with improved imaging based on modified microprinting and air-assisted deformation.

Natural compound eyes inspire the development of artificial optical devices that feature a large field of view and fast motion detection. However, the imaging of artificial compound eyes dramatically depends on many microlenses. The single focal length of the microlens array significantly limits the actual applications of artificial optical devices, like distinguishing objects at different distances. In this study, a curved artificial compound eye for a microlens array with different focal lengths was fabricated by inkjet printing and air-assisted deformation. By adjusting the space of the microlens array, secondary microlenses were created between intervals of the primary microlens. The diameter/height of the primary and secondary microlens arrays are 75/25µm and 30/9µm, respectively. The planar-distributed microlens array was transformed into a curved configuration using air-assisted deformation. Compared with adjusting the curved base to distinguish objects at different distances, the reported technique features simplicity and is easy to operate. The applied air pressure can be used to tune the field of view of the artificial compound eye. The microlens arrays with different focal lengths could distinguish the objects at different distances without additional components. When the external objects move a small distance, they can be detected by the microlens arrays due to their different focal lengths. It could effectively improve the motion perception of the optical system. Moreover, the focusing and imaging performances of the fabricated artificial compound eye were further tested. The compound eye combines the advantages of monocular eyes and compound eyes, holding great potential for developing advanced optical devices with a large field of view and automatic variable-focus imaging.

Heterogeneous compound eye camera for dual-scale imaging in a large field of view.

Multi-scale imaging with large field of view is pivotal for fast motion detection and target identification. However, existing single camera systems are difficult to achieve snapshot multi-scale imaging with large field of view. To solve this problem, we propose a design method for heterogeneous compound eye, and fabricate a prototype of heterogeneous compound eye camera (HeCECam). This prototype which consists of a heterogeneous compound eye array, an optical relay system and a CMOS sensor, is capable of dual-scale imaging in large field of view (360°×141°). The heterogeneous compound eye array is composed of 31 wide-angle (WA) subeyes and 226 high-definition (HD) subeyes. An optical relay system is introduced to re-image the curved focal surface formed by the heterogeneous compound eye array on a CMOS sensor, resulting in a heterogeneous compound eye image containing dual-scale subimages. To verify the imaging characteristics of this prototype, a series of experiments, such as large field of view imaging, imaging performance, and real-world scene imaging, were conducted. The experiment results show that this prototype can achieve dual-scale imaging in large field of view and has excellent imaging performance. This makes the HeCECam has great potential for UAV navigation, wide-area surveillance, and location tracking, and paves the way for the practical use of bio-inspired compound eye cameras.

Real-time motion detection with candidate masks and region growing for moving cameras

A fast and reliable motion detection method is proposed for moving cameras. Existing methods generally build background models and then apply background subtraction for moving pixels detection. In background modeling, statistical methods and/or features, such as edges, local binary patterns are used. Parameters of the proposed methods need to be updated adaptively for better performance. In recent studies, adaptive learning rates are used with main and candidate background models. Simple and adaptive thresholding methods are used for foreground detection in different studies. We build a statistical background model with hue–saturation–value color space and perform background subtraction with the contribution of optical flow vectors. Dense optical flow is applied between warped the background model and the current frame to estimate optical flow vectors. If a pixel has a larger magnitude, we set a weight value for each pixel in the background subtraction process. Then, we extract two candidate foreground masks and apply some postprocessing techniques to determine the final mask. Our proposed method works fast in real time, and the experimental results on two different datasets show that our approach has the highest performance compared with the methods in the literature.

Design and Implementation of Artificial Intelligence Image Detection System Based on Noise Monitoring

Image quality detection technology can replace manual inspections to automatically detect video quality, accurately analyze, judge and alarm abnormalities in video images in the surveillance system to ensure the normal operation of the expanding network video surveillance system. The thesis designs an intelligent traffic image sensor system to realize fast motion detection and object recognition of the surveillance scene. This system introduces a background model based on region segmentation and a feature-based object recognition algorithm. Experimental results show that the system can perform automatic movement detection and object classification and recognition efficiently in real time.

Deep Learning-Driven Gaussian Modeling and Improved Motion Detection Algorithm of the Three-Frame Difference Method

To enhance the effect of motion detection, a Gaussian modeling algorithm is proposed to fix holes and breaks caused by the conventional frame difference method. The proposed algorithm uses an improved three-frame difference method. A three-frame image sequence with one frame interval is selected for pairwise difference calculation. The logical “OR” operation is used to achieve fast motion detection and to reduce voids and fractures. The Gaussian algorithm establishes an adaptive learning model to make the size and contour of the motion detection more accurate. The motion extracted by the improved three-frame difference method and Gaussian model is logically summed to obtain the final motion foreground picture. Moreover, a moving target detection method, based on the U-Net deep learning network, is proposed to reduce the dependency of deep learning on the number of training datasets. It helps the algorithm to train models on small datasets. Next, it calculates the ratio of the number of positive and negative samples in the dataset and uses the reciprocal of the ratio as the sample weight to deal with the imbalance of positive and negative samples. Finally, a threshold is set to predict the results for obtaining the moving object detection accuracy. Experimental results show that the algorithm can suppress the generation and rupture of holes and reduce the noise. Also, it can quickly and accurately detect movement to meet the design requirements.

The noise navigator for MRI-guided radiotherapy: an independent method to detect physiological motion

Motion is problematic during radiotherapy as it could lead to potential underdosage of the tumor, and/or overdosage in organs-at-risk. A solution is adaptive radiotherapy guided by magnetic resonance imaging (MRI). MRI allows for imaging of target volumes and organs-at-risk before and during treatment delivery with superb soft tissue contrast in any desired orientation, enabling motion management by means of (real-time) adaptive radiotherapy.The noise navigator, which is independent of the MR signal, could serve as a secondary motion detection method in synergy with MR imaging. The feasibility of respiratory motion detection by means of the noise navigator was demonstrated previously. Furthermore, from electromagnetic simulations we know that the noise navigator is sensitive to tissue displacement and thus could in principle be used for the detection of various types of motion.In this study we demonstrate the detection of various types of motion for three anatomical use cases of MRI-guided radiotherapy, i.e. torso (bulk movement and variable breathing), head-and-neck (swallowing) and cardiac. Furthermore, it is shown that the noise navigator can detect bulk movement, variable breathing and swallowing on a hybrid 1.5 T MRI-linac system. Cardiac activity detection through the noise navigator seems feasible in an MRI-guided radiotherapy setting, but needs further optimization. The noise navigator is a versatile and fast (millisecond temporal resolution) motion detection method independent of MR signal that could serve as an independent verification method to detect the occurrence of motion in synergy with real-time MRI-guided radiotherapy.

Combined Study of a Significant Mine Collapse Based on Seismological and Geodetic Data—29 January 2019, Rudna Mine, Poland

On 29 January 2019, the collapse of a mine roof resulted in a significant surface deformation and generated a tremor with a magnitude of 4.6 in Rudna Mine, Poland. This study combines the seismological and geodetic monitoring of the event. Data from local and regional seismological networks were used to estimate the mechanism of the source and the ground motion caused by the earthquake. Global Navigation Satellite System data, collected at 10 Hz, and processed as a long-term time-series of daily coordinates solutions and short-term high frequency oscillations, are in good agreement with the seismological outputs, having detected several more tremors. The range and dynamics of the deformed surface area were monitored using satellite radar techniques for slow and fast motion detection. The radar data revealed that a 2-km2 area was affected in the six days after the collapse and that there was an increase in the post-event rate of subsidence.

COMPU-EYE: a high resolution computational compound eye

In nature, the compound eyes of arthropods have evolved towards a wide field of view (FOV), infinite depth of field and fast motion detection. However, compound eyes have inferior resolution when compared with the camera-type eyes of vertebrates, owing to inherent structural constraints such as the optical performance and the number of ommatidia. For resolution improvements, in this paper, we propose COMPUtational compound EYE (COMPU-EYE), a new design that increases acceptance angles and uses a modern digital signal processing (DSP) technique. We demonstrate that the proposed COMPU-EYE provides at least a four-fold improvement in resolution.

Functional Specialization of Neural Input Elements to the Drosophila ON Motion Detector

Detecting the direction of visual movement is fundamental for every sighted animal in order to navigate, avoid predators, or detect conspecifics. Algorithmic models of correlation-type motion detectors describe the underlying computation remarkably well. They consist of two spatially separated input lines that are asymmetrically filtered in time and then interact in a nonlinear way. However, the cellular implementation of this computation remains elusive. Recent connectomic data of the Drosophila optic lobe has suggested a neural circuit for the detection of moving bright edges (ON motion) with medulla cells Mi1 and Tm3 providing spatially offset input to direction-selective T4 cells, thereby forming the two input lines of a motion detector. Electrophysiological characterization of Mi1 and Tm3 revealed different temporal filtering properties and proposed them to correspond to the delayed and direct input, respectively. Here, we test this hypothesis by silencing either Mi1 or Tm3 cells and using electrophysiological recordings and behavioral responses of flies as a readout. We show that Mi1 is a necessary element of the ON pathway under all stimulus conditions. In contrast, Tm3 is specifically required only for the detection of fast ON motion in the preferred direction. We thereby provide first functional evidence that Mi1 and Tm3 are key elements of the ON pathway and uncover an unexpected functional specialization of these two cell types. Our results thus require an elaboration of the currently prevailing model for ON motion detection and highlight the importance of functional studies for neural circuit breaking.

Fast-gesture recognition and classification using Kinect: an application for a virtual reality drumkit

In this paper, we present a system for the detection of fast gestural motion by using a linear predictor of hand movements. We also use the proposed detection scheme for the implementation of a virtual drumkit simulator. A database of drum-hitting motions is gathered and two different sets of features are proposed to discriminate different drum-hitting gestures. The two feature sets are related to observations of different nature: the trajectory of the hand and the pose of the arm. These two sets are used to train classifier models using a variety of machine learning techniques in order to analyse which features and machine learning techniques are more suitable for our classification task. Finally, the system has been validated by means of the Kinect application implemented and the participation of 12 different subjects for the experimental performance evaluation. Results showed a successful discrimination rate higher than 95 % for six different gestures per hand and good user experience.

Temporal analysis for fast motion detection in a crowd

We present a fast motion detection technique in a crowd as an abnormal motion based on optical flow and a motion history image (MHI). Since a camera view is usually not in perpendicular with motion direction, the velocity of motion is not uniform spatially. Instead of object detection directly from an image, we separate an image into several blocks. In this paper, we propose a novel method to analyze a motion using MHI representation, called a shift space and a shift histogram. Together with a velocity histogram, the method can detect fast motion in a crowd, realizing local abnormal event detection. The performance of the proposed method is experimentally illustrated and evaluated.

Fabrication of curved microlens array using a drop-on-demand droplet generator and polydimethylsiloxane replica mold

A simple approach was demonstrated to fabricate the curved microlens array (MLA) using a drop-on-demand (DOD) droplet generator and a polydimethylsiloxane (PDMS) replica mold. The planar microlens array (planar MLA) was fabricated using a simple and cost-effective DOD droplet generator, and then the curved MLA was fabricated by multiple replication processes. The PDMS elastomer membrane mold replicated from the planar MLA was used to transfer the microlenses from a planar substrate onto a spherical surface. The NORLAND65 (NOA65) optical adhesive was chosen to be the material of the curved MLA due to its good optical properties. The curved MLA, whose radii of curvature of the spherical surface were approximately 5.1 and 1.8 mm, was fabricated. The morphology damage during the replication process was avoided by the surface treatment of the PDMS elastomer membrane mold and the deformation of the curved microlens was analyzed and was ranged from 2.1% to 8.5%. The surface quality of the curved microlenses was measured by a white light interferometer and an atomic force microscope and the results showed that the arithmetical mean deviation of the profile R a was 0.54 nm with a 2×2- μ m 2 scanned area, while the resultant surface showed an excellent surface smoothness. The imaging and focus performance of the curved MLA were measured by a projection experiment. The experimental results demonstrated a high potential for curved MLA being applied to scale-invariant processing, robot vision, and fast motion detection.

Neural Mechanism for Sensing Fast Motion in Dim Light

Luminance is a fundamental property of visual scenes. A population of neurons in primary visual cortex (V1) is sensitive to uniform luminance. In natural vision, however, the retinal image often changes rapidly. Consequently the luminance signals visual cells receive are transiently varying. How V1 neurons respond to such luminance changes is unknown. By applying large static uniform stimuli or grating stimuli altering at 25 Hz that resemble the rapid luminance changes in the environment, we show that approximately 40% V1 cells responded to rapid luminance changes of uniform stimuli. Most of them strongly preferred luminance decrements. Importantly, when tested with drifting gratings, the preferred speeds of these cells were significantly higher than cells responsive to static grating stimuli but not to uniform stimuli. This responsiveness can be accounted for by the preferences for low spatial frequencies and high temporal frequencies. These luminance-sensitive cells subserve the detection of fast motion under the conditions of dim illumination.

Biologically Inspired CMOS Image Sensor for Fast Motion and Polarization Detection

A complementary-metal-oxide semiconductor (CMOS) image sensor replicating the perception of vision in insects is presented for machine vision applications. The sensor is equipped with in-pixel analog and digital memories that allow in-pixel binarization in real time. The binary output of the pixel tries to replicate the flickering effect of an insect's eye to detect the smallest possible motion based on the change in state of each pixel. The pixel level optical flow generation reduces the need for digital hardware and simplifies the process of motion detection. A built-in counter counts the changes in states for each row to estimate the direction of the motion. The designed image sensor can also sense polarization information in real time using a metallic wire grid micropolarizer. An extinction ratio of 7.7 is achieved. The 1-D binary optical flow is shown to vary with the polarization angle of the incoming light ray. The image sensor consists of an array of 128 × 128 pixels, occupies an area of 5 × 4 mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and it is designed and fabricated in a 180-nm CMOS process.

Real-time Alert of Motion Detection Using Microcontroller

This paper addresses the problem of motion detection and moving objects of variable appearance in challenging scenes rich with features and texture. It is made particularly difficult by the nature of objects encountered in such scenes. The research included a method which uses fast motion detection and segmentation as constraint of differences between frames of video, and when changes in each pixel of an image, will alert start and message displayed on LCD that plugged to microcontroller.

Fast motion detection from airborne videos using graphics processing unit

DCM Research Resources LLC, Germantown, Maryland 20874Abstract. In our previous work, we proposed a joint optical flow and principal componentanalysis (PCA) approach to improve the performance of optical flow based detection, wherePCA is applied on the calculated two-dimensional optical flow image, and motion detectionis accomplished by a metric derived from the two eigenvalues. To reduce the computationaltime when processing airborne videos, parallel computing using graphic processing unit(GPU) is implemented on NVIDIA GeForce GTX480. Experimental results demonstrate thatour approach can efficiently improve detection performance even with dynamic background,and processing time can be greatly reduced with parallel computing on GPU.

A simple route to fabricate artificial compound eye structures

A biologically inspired compound-eye structure, which composes of ~5,867 honeycomb-patterned microlenses, was fabricated on a hemispherical shell. The fabrication process was simple and low-cost, which involves a femtosecond laser-enhanced wet etching and casting process followed by a thermomechanical process to convert the film into a hemispherical surface. By optimizing the parameters of thermomechanical process to form the curvilinear surface, the experimental result shows that the microlenses are omnidirectionally aligned on the dome with lens diameters of ~85 µm and the angle between two lens of ~2°, and the individual microlenses have rudimentary focusing and imaging properties. The artificial compound-eye structure fabricated by this method has great potential applications in scale-invariant processing, robot vision, and fast motion detection.

Robustness of optimal intermittent search strategies in one, two, and three dimensions

Search problems at various scales involve a searcher, be it a molecule before reaction or a foraging animal, which performs an intermittent motion. Here we analyze a generic model based on such type of intermittent motion, in which the searcher alternates phases of slow motion allowing detection and phases of fast motion without detection. We present full and systematic results for different modeling hypotheses of the detection mechanism in space in one, two, and three dimensions. Our study completes and extends the results of our recent letter [Loverdo, Nat. Phys. 4, 134 (2008)] and gives the necessary calculation details. In addition, another modeling of the detection case is presented. We show that the mean target detection time can be minimized as a function of the mean duration of each phase in one, two, and three dimensions. Importantly, this optimal strategy does not depend on the details of the modeling of the slow detection phase, which shows the robustness of our results. We believe that this systematic analysis can be used as a basis to study quantitatively various real search problems involving intermittent behaviors.

Microfabricated ommatidia using a laser induced self-writing process for high resolution artificial compound eye optical systems

A microfabricated compound eye, comparable to a natural compound eye shows a spherical arrangement of integrated optical units called artificial ommatidia. Each consists of a self-aligned microlens and waveguide. The increase of waveguide length is imperative to obtain high resolution images through an artificial compound eye for wide field-of - view imaging as well as fast motion detection. This work presents an effective method for increasing the waveguide length of artificial ommatidium using a laser induced self-writing process in a photosensitive polymer resin. The numerical and experimental results show the uniform formation of waveguides and the increment of waveguide length over 850 microm.