Top-view Camera Research Articles

Multiple View Surveillance using Image Registration

In the area of activity monitoring, video surveillance was mainly used for monitoring the scene by security personnel. However, due to high increase of unethical activities it is being strongly considered as a must for vigilance authorities, even sometimes for some unique purposes such as exam surveillance. The main theme is to ensure that students take their examinations the right way instead of indulging in unfair methods or malpractices. The proposed system aims to detect suspicious activities which occur in the classroom, thereby ruling out copying. We aim to develop a system that will identify the students from the top view camera using the image registration technique and check to see if he/she indulges in any form of malpractices or suspicious activities during the entire course of an examination. To get the aimed results the whole system has been divided into two modules through which the suspicious activities are classified from the normal activities. The first one includes the image registration technique which is used as a novel approach to determine the position of the student from the top view camera. The second module proposes the usage of object detection and skin detection. Finally we compare the normal frames and the suspicious frames based on their respective threshold values which are generated using the threshold technique. Also the accuracy of the system is checked using SVM classification.

Non-invasive devices and methods for large animal monitoring using automated video processing

ObjectivesIt is now standard for polysomnographical equipment to include video recording, although this modality is generally underexploited, since there is no automated processing associated with the latter. In the present report, we investigated the set of features that can be automatically extracted from a video recording, in the context of monitoring of freely moving, non-sedated, newborn lambs. Material and methodsOur database contained seven lambs and a total of 11 recordings, using two different cameras allowing a top view and a side view. Using appropriate methodologies, we show that it is possible to estimate the lamb's movements, its posture (standing or lying) as well as its covered trajectory. ResultsResults are discussed as a function of the camera and show that side view recording is well suited for accurate scoring of the lamb's posture, whereas trajectory is best estimated using the top view camera. On the other hand, both cameras provide qualitatively similar results for the estimation of movement of the animals. ConclusionThe data gained from automated video processing, as reported herein, may have multiple applications, especially for animal studies, but may also be extended to human sleep monitoring.

Analysis of aggressive behaviours of pigs by automatic video recordings

Aggression among pigs in today’s production systems results in negative impact on health and welfare of animals as well as on productivity of the systems. Precision Livestock Farming technology might potentially offer a possibility to monitor and reduce the level of aggression and hence its negative impact. This paper reports about the initial part of a larger study investigating the possibilities of applying continuous automatic monitoring of aggressive behaviour among pigs. It investigates how behavioural patterns in pig’s aggressive behaviour can be identified and utilized in order to predict severe forms of aggression (biting) expressed in later phases of aggressive interactions.An experiment was carried out at a commercial farm on a group of 11 male pigs weighing on average 23kg and kept in a pen of 4m×2.5m. During the first 3days after mixing in total 8h of video recording were registered with a top view camera for later analysis of animal behaviour. As a result of labelling of the video recordings, 157 aggressive interactions were identified with 12 behaviour types expressed for 860 times within the interactions. The identified interactions were divided into interactions that led to biting and those that did not lead to biting behaviour. The interactions that led to biting behaviour accounted for 36.3% (57) of all aggressive interactions while interactions that did not lead to biting behaviour were 63.7% (100) of the interactions. The average duration of initiating (nosing) phase of aggressive interactions (3.32s) lasted longer (P<0.05) in interactions that led to biting behaviour than in interactions that did not lead to biting behaviour (1.94s). The next phase of aggressive interactions – medium phase – similarly to initiating phase, lasted on average longer (18.21s) (P<0.01) in interactions that led to biting behaviour than in interactions that did not lead to biting behaviour (16.15s). With the differences found between interactions that led and did not lead to biting behaviour it seems to be possible to discriminate between both types of interactions in an early phase of aggression. The differences found might serve as early signs in a management support system that aims to prevent severe forms of aggressive behaviour (biting) among pigs.

An Efficient and Effective Localization Method for Networked Disjoint Top-View Cameras

Disjoint top-view networked cameras are among the most commonly utilized networks in many applications. One of the open questions for these cameras' study is the computation of extrinsic parameters (positions and orientations), named extrinsic calibration or localization of cameras. Current approaches either rely on strict assumptions of the object motion for accurate results or fail to provide results of high accuracy without the requirement of the object motion. To address these shortcomings, we present a location-constrained maximum a posteriori (LMAP) approach by applying known locations in the surveillance area, some of which would be passed by the object opportunistically. The LMAP approach formulates the problem as a joint inference of the extrinsic parameters and object trajectory based on the cameras' observations and the known locations. In addition, a new task-oriented evaluation metric, named MABR (the Maximum value of All image points' Back-projected localization errors' L2 norms Relative to the area of field of view), is presented to assess the quality of the calibration results in an indoor object tracking context. Finally, results herein demonstrate the superior performance of the proposed method over the state-of-the-art algorithm based on the presented MABR and classical evaluation metric in simulations and real experiments.

Development of an early warning system for a broiler house using computer vision

Anomalous animal behaviour and reduced growth rate are just a few signs that can indicate an undesired situation in a broiler house. It is important that problems such as malfunctioning feeding and drinking lines are detected at an early stage to avoid a negative impact on the welfare or the production of broilers. This paper introduces an automated method to detect problems in a broiler house using cameras and image analysis software. In an experiment with Ross 308 broilers conducted in a commercial broiler house with 28,000 animals, three top-view cameras were mounted at a 5 m height and continuously monitored a floor space of 19.8 by 63.5 m. Analysis software then translated these images into an animal distribution index. Animal distribution index is known to be related to welfare and it can be affected by equipment malfunctioning in a broiler house. Thus, the final objective was to develop a system that could report malfunctioning in a broiler house to the farmer in real-time. Based on these data, a linear real-time model was developed and tested to model the animal distribution index in response to light input. Using this model, the animal distribution index could be predicted online. Comparing these predicted values with the real-time measurements makes it possible to detect any malfunctioning. Results showed that this method could report 95.24% of events (20 out of 21) in real-time, demonstrating a high potential of using automatic monitor tools for broiler production over a complete growing period.

Influence of evaporation rate on cracks’ formation of a drying drop of whole blood

The influence of the drying rate on pattern formation of a dried drop of whole blood is investigated. A range of relative humidity (RH) between 12.0% and 66.5% is studied. Drops of whole blood of same volume are gently deposited on microscope glass substrates. A top-view camera operates at a typical frame rate of 10 frames per second, which is sufficient to monitor the drying regime of the drop (deposition, gelation, and fracturation). Also, the evolution of the morphological structure of patterns made up of cracks and their arrangement in the corona region is studied. For this purpose, a manual segmentation is used as a reference method for the validation of our automatic developed segmentation method. This automatic segmentation method enables automatic extraction and morphological characterization of patterns. We show that the evaporation rate influences structural distribution of plaques, and consequently, the final plaques diameter at the end of the evaporation process. At low drying rate (high RH), plaques diameter show good agreement between automatically and manually segmented plaques with a slight difference. Also, automatic segmentation overestimates angular deviation. Our experimental results also show that the length of the corona region remains constant whatever is the RH levels.

Multivariate dice recognition using invariant features

A system is proposed for automatic reading of the number of dots on dice in general table game settings. Different from previous dice recognition systems that recognize dice of a specific color using a single top-view camera in an enclosure with controlled settings, the proposed one uses multiple cameras to recognize dice of various colors and under uncontrolled conditions. It is composed of three modules. Module-1 locates the dice using the gradient-conditioned color segmentation, proposed, to segment dice of arbitrary colors from the background. Module-2 exploits the local invariant features good for building homographies, giving a solution to segment the top faces of the dice. To identify the dots on the segmented top faces, a maximally stable extremal region detector is embedded in module-3 for its consistency in locating the dot region. Experiments show that the proposed system performs satisfactorily in various test conditions.

Influence of relative humidity on spreading, pattern formation and adhesion of a drying drop of whole blood

The drying of a drop deposited on a solid substrate has been the subject of several investigations including biomedical and forensic fields. These studies also focus on the complex final drop pattern observed at the end of the evaporation process. Our experimental work aims to investigate the effect of relative humidity (RH) on the spreading behavior and on the pattern formation of a dried drop of whole blood at the end of the evaporation process. A range of RH between 13.5% and 78.0% is studied. Drops of blood of same volume are gently deposited on ultraclean microscope glass substrates. A top-view camera allows for the monitoring of the drying regime (deposition, gelation and fracturation). We show that RH influences the contact angle, and the final wetting diameter and consequently, the final deposition pattern at the end of the evaporation process. A good agreement has been observed between our results for whole blood and the experimental work of Chhasatia et al. [Appl. Phys. Lett. 97, 231909 (2010)] performed an aqueous drop of 1.1μm colloids. Our experimental measurements are in a good agreement with the purely diffusive model where the wetting diameter and the contact angle are function of RH. Our experimental results also show that the transition between the purely convective evaporation phase and the gelation phase occurs always at 65% of the total drying time whatever is the RH levels.

Path planning for micro-part assembly by using active stereo vision with a rotational mirror

Our proposed active vision system for micro assembly is equipped with a top view camera and a side view camera with a rotational mirror system, such that the viewing direction can be changed freely. The rotational side view helps minimize an occlusion by changing the view direction and measuring three-dimensional (3-D) information using stereo reconstruction. We investigate the global path planning problem, which considers obstacles and occlusions on the image space of each camera simultaneously. General 3-D space-based path planning algorithms may fail in terms of successful safe path generation due to camera or system calibration uncertainty. The proposed approach generates a safe path on two image spaces using a fast marching method, and follows this path based on a visual tracking method in the image space. In order to avoid a visual occlusion and obstacle, the travel cost is built by using a potential field method. For efficient path generation in image spaces, epipolar boundary constraints are considered in the potential field generation. Simulation and experimental results show the feasibility of the proposed algorithm for the assembly of microelectromechanical system parts.

2-Dimensional Wavelet pre-processing to extract IC-Pin information for disarrangement analysis

Due to higher processing power to cost ratio, it is now possible to replace the manual detection methods used in the IC (Integrated Circuit) industry by Image-processing based automated methods, to detect a broken pin of an IC connected on a PCB during manufacturing, which will make the process faster, easier and cheaper. In this paper an accurate and fast automatic detection method is used where the top view camera shots of PCBs are processed using advanced methods of 2-dimensional discrete wavelet pre-processing before applying edge-detection. Comparison with conventional edge detection methods such as Sobel, Prewitt and Canny edge detection without 2-D DWT is also performed.

Privacy-preserved behavior analysis and fall detection by an infrared ceiling sensor network.

An infrared ceiling sensor network system is reported in this study to realize behavior analysis and fall detection of a single person in the home environment. The sensors output multiple binary sequences from which we know the existence/non-existence of persons under the sensors. The short duration averages of the binary responses are shown to be able to be regarded as pixel values of a top-view camera, but more advantageous in the sense of preserving privacy. Using the “pixel values” as features, support vector machine classifiers succeeded in recognizing eight activities (walking, reading, etc.) performed by five subjects at an average recognition rate of 80.65%. In addition, we proposed a martingale framework for detecting falls in this system. The experimental results showed that we attained the best performance of 95.14% (F1 value), the FAR of 7.5% and the FRR of 2.0%. This accuracy is not sufficient in general but surprisingly high with such low-level information. In summary, it is shown that this system has the potential to be used in the home environment to provide personalized services and to detect abnormalities of elders who live alone.

Influence of Substrate Nature on the Evaporation of a Sessile Drop of Blood

We fully characterize the natural evaporation of human drops of blood from substrates and substrate-dependent behavior. The heat flux adsorbed by the drops for evaporation is measured by means of a heat flux meter. A side-view measurement enables access to the drop contact angle, wetting diameter, and initial height. A top-view camera allows for the monitoring of the drying regime (deposition, gelation, and fracturation). This directly measured heat flux is related to the evaporative mass flux obtained from the mass of the drop, and the two show good agreement. Both types of measurements indicate that regardless of the substrate type, there is first a linearly decreasing regime of evaporation when the drop is mostly liquid and a second regime characterized by a sharp decrease. We show that the evaporation dynamics are influenced by the substrate’s wettability but not by the substrate’s thermal diffusivity. The different regimes of evaporation exhibited by glass and metallic substrates are explained in terms of evaporation fluxes at the drop surface. In the case of wetting drops (below 40 deg), the evaporation flux is very important along the drop periphery and decreases across the interface, whereas in the case of nonwetting drops (about 90 deg), the evaporation flux is almost uniform across the droplet’s surface. We show that these different evaporation fluxes strongly influence the drying behavior. In the case of metallic substrates, this enables the formation of a uniform "glassy skin" around the droplet surface and, in the case of glass substrates, the formation a skin along the drop periphery with an inward gelation front. This behavior is analyzed in terms of the competition between the drying time and the gel formation time. Unstable drop surfaces were observed at high initial contact angles and are very similar to those of polymer drops.

Pedestrian Detection for Counting Applications Using a Top-View Camera

This paper presents a pedestrian detection framework using a top-view camera. The paper contains two novel contributions for the pedestrian detection task: 1. Using shape context method to estimate the pedestrian directions and normalizing the pedestrian regions. 2. Based on the locations of the extracted head candidates, system chooses the most adaptive classifier from several classifiers automatically. Our proposed methods may solve the difficulties on top-view pedestrian detection field. Experimental was performed on video sequences with different illumination and crowed conditions, the experimental results demonstrate the efficiency of our algorithm.

Design of a body-driven multiplayer game system

We have designed and evaluated a novel multiplayer game system using just one top-view camera. With the proposed system, player avatar movement can be directly mapped to the physical movement of the player, accompanied by additional hand gestures triggering more complex actions. This article presents a study of the concepts of body-driven multiplayer games using the proposed system. We have created four different test games using human-centered design (HCD). We describe both the computer vision- based implementation and the lessons we learned about designing effective content for interactive body-driven multiplayer games.