To Derive An Effective Algorithm That Can Be Used For Detecting Motion In Computer Based Security Surveillance Cameras
ABSTRACTDigital data has been grows exponentially as observed from recent years. We are at a time where computing power, server space, video capture equipment and networking technology have sufficiently expanded to make remarkable influences in motion detection, event detection and situation awareness. The most viable aspect of these developments is the capability to browse over large amounts of stored data. Most prior technologies have depended on text such as those mined from closed captioning, speech analysis or even manual annotation, we would love to search based on automated recognition of visual events and motion in video surveillance. Situation awareness is the key to security and therefore requires information collected over a wide scope of time and space. Smart video surveillance systems are able to enhance situational awareness over a wide spectrum. However, at the moment, the technology component is dynamic in isolation. This paper sets out to establish the current status of motion detection, event detection and situation awareness. It goes further to explain the different techniques used in motion detection, and the most current and effective approaches used in event detection, motion detection and situation awareness . This research aims at developing a new way of quality motion detection using high resolution.
DEFINITION OF TERMSCCD-Charge Coupled Device
CCTV –Closed Circuit Television
FEMD- Front end Molecular DymanicsOnSSI- On-Net Surveillance Systems
Ocularis – a security surveillance system adapted by many firms currently
SCAP- Security Content Automated Protocol
BriefCam- an IP video surveillance production firm
SAGAT- Situation awareness global assessment technique
SME- subject matter experts
SA-Situation Awareness
SPAM- situation present assessment method
SART- situation awareness rating technique
SARS- Situation awareness rating scales technique
SME-Subject Matter Experts
MARS- Mission Awareness Rating Scale
SABARS- Situation Awareness Behavioral Rating Scale
MRF- Markov Random Fields
HMM- Hidden Markov Model
GMM- Gaussian mixture model
MAP- Maximum APosterioriEM- Expectation-Maximization
V.S-Video Synopsis – a technique for reducing long video footage into a sherter summary
Table of Contents
TOC o “1-3” h z u ABSTRACT PAGEREF _Toc338846493 h iDEFINITION OF TERMS PAGEREF _Toc338846494 h iiCHAPTER 1 PAGEREF _Toc338846495 h 1INTRODUCTION PAGEREF _Toc338846496 h 11.1 INTRODUCTION PAGEREF _Toc338846497 h 11.2 STATEMENT OF THE PROBLEM PAGEREF _Toc338846498 h 11.3 BACKGROUND OF THE PROBLEM PAGEREF _Toc338846499 h 21.4 OBJECTIVES PAGEREF _Toc338846500 h 41.5 CURRENT TECHNIQUES OF EVENT DETECTION PAGEREF _Toc338846501 h 4(a) Shape Matching PAGEREF _Toc338846502 h 4(b) Flow Matching PAGEREF _Toc338846503 h 5(c) Space-time Interest Points PAGEREF _Toc338846504 h 5(d) Pose Tracking PAGEREF _Toc338846505 h 6(e) 3D Chamfer Matching PAGEREF _Toc338846506 h 6CHAPTER 2 PAGEREF _Toc338846507 h 8METHODOLOGY PAGEREF _Toc338846508 h 82.1 SITUATIONAL AWARENESS TECHNIQUES PAGEREF _Toc338846509 h 82.1.1 Situational Awareness requirements analysis PAGEREF _Toc338846510 h 82.1.2 Freeze probe techniques PAGEREF _Toc338846511 h 82.1.3 Real-time probe techniques PAGEREF _Toc338846512 h 92.1.4 Self-rating techniques PAGEREF _Toc338846513 h 102.1.5 Observer rating techniques PAGEREF _Toc338846514 h 122.2 PERFORMANCE MEASURES PAGEREF _Toc338846515 h 122.3 PROCESS INDICES PAGEREF _Toc338846516 h 122.4 CURRENT TECHNIQUES FOR MOTION DETECTION PAGEREF _Toc338846517 h 132.4.1 Optical flow PAGEREF _Toc338846518 h 13Methods for determining optical flow PAGEREF _Toc338846519 h 142.6.Bayesian Illumination-Invariant Motion Detection PAGEREF _Toc338846520 h 152.6.1 Markov Random Fields (MRFs) as a priori models PAGEREF _Toc338846521 h 152.6.2 Hidden Markov model PAGEREF _Toc338846522 h 152.6.3 Gaussian mixture model PAGEREF _Toc338846523 h 17Results from the objectives PAGEREF _Toc338846524 h 19CHAPTER THREE PAGEREF _Toc338846525 h 20SYSTEM ANALYSIS PAGEREF _Toc338846526 h 203.1 SYSTEM ANALYSIS PAGEREF _Toc338846527 h 203.1.1Requirements specification PAGEREF _Toc338846528 h 203.1.2 System analysis and design stages PAGEREF _Toc338846529 h 20CHAPTER FOUR PAGEREF _Toc338846530 h 23SYSTEM DESIGN PAGEREF _Toc338846531 h 234.1 PROGRAM DESIGN PAGEREF _Toc338846532 h 234.2 DESIGN APPROACHES. PAGEREF _Toc338846533 h 244.3 SYSTEM DIAGRAM/DESIGN TOOL PAGEREF _Toc338846534 h 24(a)Flowchart PAGEREF _Toc338846535 h 24(b)Decision tree PAGEREF _Toc338846536 h 25(c)Pseudo code PAGEREF _Toc338846537 h 264.4 OUTPUT DESIGN PAGEREF _Toc338846538 h 27CHAPTER FIVE PAGEREF _Toc338846539 h 27SYSTEM DOCUMENTATION PAGEREF _Toc338846540 h 275.1 REQUIREMENT SPECIFICATIONS PAGEREF _Toc338846541 h 275.1.1 Program Requirements. PAGEREF _Toc338846542 h 27Hardware: PAGEREF _Toc338846543 h 285.2 PROGRAM CODING, OPERATION AND FUNCTIONALITY: PAGEREF _Toc338846544 h 295.2.1 programming language choice PAGEREF _Toc338846545 h 295.2.2 Program operation PAGEREF _Toc338846546 h 29
CHAPTER 1INTRODUCTION1.1 INTRODUCTIONComputer vision majorly focuses on problems that involve integrating computers with the surrounding environment through visual means. An example is object recognition which involves a computer sensing the presence of a known object in an image, given some information about the real picture perception of the object (Gnawali, Jang and Paek).
Enforcing greater standards of security at public access areas such as bus terminals, railway stations, airports and seaports has recently become a vital aspect. Various technologies have been implemented to increase security including personnel and object systems, security organs for tracking, biometric identification systems and video surveillance systems to monitor motion and events. The present surveillance systems are more of wide scale video recorders. The main aim of event detection is to sort and localize specific spatio-temporal patterns in video, such as a person waving his/her hand. As observed by Ke et al. (14) and Shechtman &Iran (24), the task is similar to object detection in various ways as the pattern can be located anywhere in the scene and requires reliable detection in the presence of significant background clutter. In a bid to help enforce security controls various techniques to motion detection, event detection and situation awareness have been enforced (Ke, Sukthankar and Hebert).
1.2 STATEMENT OF THE PROBLEMCurrently event detection has become more sophisticated with recent tools being put in place to analyze the ever rising volumes of video from various sources such as video surveillance from closed circuit cameras (CCTV’s) . The long hours of video footage has to be analyzed for security purposes by security personnel, which usually requires long man-hours and is prone to errors and personal judgment. There is a need to develop computer software that can do the same analysis more efficiently and precisely in real-time and provide feedback to the management.
1.3 BACKGROUND OF THE PROBLEMToday’s motion detecting cameras all make use of variations of the same technique for detecting scene displacements. Ideally, each successive picture is mapped to a reference image pixel by pixel. In case the number of pixels that do not resemble goes beyond some preset threshold, the picture is interpreted by the computer as different from the reference image and therefore reported as anomaly or motion is detected and the alarm is set. Normally thisrequires considerable picture preprocessing to minimize random noise from dark currents of CCD and other sources.
An easy conceptual evaluation of the FEMD process is:
Is the pixel by pixel separation between the picture at hand and the reference image, the factor r j is a noise minimization term, and e is the sensitivity threshold. There exist many different versions of this process that have been implemented. However, it is not only motion that can cause sufficient disparities to generate an alarm. The camera also detects and records all scene movements; a lot of which are incidental to the real surveillance and do not have safeguard impacts (Ke, Sukthankar and Hebert). Nevertheless, in spite of these limitations, today’sFEMD technology has significantly minimized the sizes of data that must be gathered and reviewed as opposed to traditional time-lapse surveillance techniques. Smart Videos attempt to minimize both false alarms and incidental alarms by widening the FEMD process to accommodate changes to the geometry of the field of perception; and then employ the geometry information to sort and identify moving objects (Kulkarni, Ganesan and Shenoy).
A larger part of the prior works to detect and sort objects was primarily geared towards satellite imagery]. The methods developed for this application are reliant on certain scope conditions that are found in this scenario; like fixed vertical aspect and minimal motion. There was no bid to actually detect and monitor motion in the sense of Safeguards, and the layout of the objects of interest was considerably constant. A lot of recent work has been based on texture-mapping within color-space; which is not available in imagery Safeguards. Many developments have been made in towards the aim of automatically tracking, identifying, and separating objects of interest in surveillance. A lot of work still remains incomplete to achieve this goal, and Smart Video is not expected to rise as a long term technology until the end of the century. However, the outcomes to date already give potential benefits in remote monitoring Safeguards environment today (Gnawali, Jang and Paek). The demonstrated capability to ascertain that a scene displacement was not due to lighting inconsistencies, but was indeed as a result of the movement of a discrete object in the area of view has the specific potential to eliminate or reduce the false alarm rate. In crucial complex well defined Safeguards scenarios it might be possible to discriminate between a small size, quick, vertical object and a large size, slow, horizontal object reliably (Gnawali, Jang and Paek). This would enable the system to weigh out incidental motion alarms that have no Safeguards sensible information. all of these capabilities can minimize the amount of data to be gathered and reviewed; and may provide a means of minimizing the periods an inspector should be on site at the time of activities. The major challenge in deployment of these methods is the frame-rate of now available authenticating cameras. It is projected, however, that this challenge will be easier and quick to face out than was the processing technology development (Gnawali, Jang and Paek).
Incorporation of a situational awareness and consistent tracking program is becoming a basic function and, in some scenarios a must, for IT Security organizations currently. “Situational awareness” is defined by a Chief Information Security Officer (CSO) working group as, “is having access to near real-time information that accurately describes the security posture of the organization with drill down capability to a single actor, system or application. Information includes sufficient knowledge about: threats, vulnerabilities, resources/assets, and the significance of the interactions between the three components” (Governance and Risk Solutions (2011)To attain a strategic level of security risk management in enterprises, a top-down integration of existing security features and viable datasets is essential (Nam, Abe and Kamwata). The relation and analysis of that information provides insight into an enterprises security risk status. Awareness is aired by alerts, summary, metrics, details, and narrative reporting present in tailored dashboard views. Many security solutions currently provide a tactical view into critical areas of the organization, centering on people, asset, or information. Risk vision relates data from prior security and IT tools merge that with enterprises, compliance, and risk information to portray a detailed and risk optimized understanding of security posture, at the business large picture level. With a wide range of out-of-the-box relations to eminent security and IT management systems and firm automation technologies relying on open standards such as SCAP, Risk vision may automate large amount of security data in real-time to develop true situational awareness intelligence (Walker).
Ocularis from On-Net Surveillance Systems (OnSSI) has been upgraded to put management of physical security data even more powerfully in the users hands (Ke, Sukthankar and Hebert). It now enables advanced event detection, shared alert management, bookmarking and transfer of evidence in less time. Ocularis Video Synopsis, is a module emanating from a merger between OnSSI and BriefCam, differently enables the compression of video activity into a short timeframes for faster identification of crucial events. Long hours of video footage can be sliced down into a 60 second video clip. Ocularis is an intelligent IP-based video surveillance control and management software solution that intensifies the purpose of the IP Video system to be fully-fledged, video-based management system for video data within the enterprise (Kulkarni, Ganesan and Shenoy)
1.4 OBJECTIVESThis paper will be guided by the following objectives:
To derive aneffective algorithm that can be used for detecting motion in computer based security surveillance cameras
To develop a computer software that can effectively detect motion using the developed algorithm.
1.5 CURRENT TECHNIQUES OF EVENT DETECTION(a) Shape MatchingThis paper will look at Shape-based techniques that treat the Spatio-temporal volume of a video sequence as a 3D object. Various video events produce unique shapes; the main aim of such goals is to identify an event by recognizing its shape. This incorporates a pool of techniques to present the shape character of an event, with the likes of shape invariants. For computational efficiency and greater robustness
To action variations, Bobick and Davis project the Spatio-temporal volume down to motion-history images, in which (Weinland et al. 3). These techniques perform ultimately when the interest action is done in a segmentation reliable enabling platform. For stationery scene specifically, background subtraction may result in quality spatio-temporal volumes which are subject to analysis. It is regrettable that conditions such as these do not apply in real world videos because of the various objects movement and scene clutter (Gnawali, Jang and Paek).
(b) Flow MatchingFlow-based techniques for event detection act on the spatio-temporal sequence directly, bids to unveil pattern specifics by brute force correlation facing out segmentation. Templates together with videos can also be correlated in a bid to recognize distanced objects.(Shechtman and Irani 16) suggest an algorithm for correlating videos without explicitly computing the optical flow against spatio-temporal event templates, which of course has limitations like noise in boundaries of objects. Recently, the SVM classifier intentionally trained on histograms of optical flow is applied to actually find tennis actions (Jhuang et al 10).on the other hand applies features biological inspired, which consist of optical flow, for action recognition (Ke, Sukthankar and Hebert).
(c) Space-time Interest PointsSpace-time interest points recently have become famous in the action recognition environments, with multiple parallels to the approach traditional interest points have been used to recognize objects. Interest point’s sparsity and their subsequent computational efficiency are convincing, but then they have limitations with those of 2D analogues, examples being, inability to capture smooth motion transitions and its trend to result to spurious detections at object boundaries (Ke, Sukthankar and Hebert).
They are dependent on bringing out the local region surrounding an interest area applying robust representations to geometric noise and perturbations, yet unique sufficiently to reliably locate the local region. However, these techniques assume that one can reliably detect a number of stable interest points in the video sequence. In space-time interest points this implies that the video sequence ought to encompass many motion critical events instances areas in which object are rapidly dynamic in its motion direction such as the duplicating path traced by a walking person’s shoe. It is unfortunate that these techniques however, are not capable of detecting useful interest points in several situations in which the motions contain no sharp extreme, they are also ignited by the appearance of shadows and highlights in the video sequence, These unstable events are critical to lighting conditions and may bring down accuracy for recognition for the action of interest (Ke, Sukthankar and Hebert).
(d) Pose TrackingMethods relying on the video frame-by-frame process tracking and object segment from background clutter, simply by relating the current frame against some model. Tracing object’s motion over some period, a trace of model parameters is produced; in which this trace can be related with that of the planned patio-temporal pattern to ascertain whether the observed event is of interest. A view-based approach is taken that does not track these model arguments. Giving room for generalization to human, animal, or mechanical actions in absence of preceding models of these objects, approaches relying on tracking integrate existing scope knowledge about the projected event in the model; the system can also accommodate web based queries due to the processing of single frames at a time. However, it’s still critical to initialize tracking models, especially if the context contains distracting objects. Recent work has however, proved enough developments in cluttered environments, tracking still remains critical in such environments, while the tracker result seems to be noisy. There is a substitute approach which is using tracking-based event detection that concentrates on multi-agent activities, in a case where each actor is tracked as a bloband activities are rated in consideration with locations observed and spatial interactions between blobs. These models sufficiently work well in expressing activities (Walker).
(e) 3D Chamfer MatchingHumans make use of extensive previous data in formof various visual cues such as color, texture and shape to recognize objects. Machine vision algorithms attempt to imitate human perception system by learning similar priors based on exemplars. However, collecting the required training data is a tedious task and significantly limits the effectiveness of these algorithms in many cases. For instance, it is much simpler to trace an image collection by a one exemplar given by a user than learning every possible object class beforehand. Still, recognition of objects in images ,using only a few exemplars remains to be a very critical problem. The most eficient method applicable to solving this problem is to make use of features that contain the least variability within object classes and across imaging conditions mingled up with a same measure that models the maximum invariance’s (Ke, Sukthankar and Hebert).
Chamfer matching image processing technique places different same patient scans automatically in same co-ordinate system emanating from a specific part of the anatomy. Chamfer matching can be applied in analyzing organ motion, tumor control follow up and verifying treatment delivery by use of electronic portal imaging and CT scans. It basically maps drawings into images. This image may accommodate another drawing or can be the output of image processing done to an original image. The merit of fit between the drawing and the image is evaluated by a function that ascertains a cost, for drawing to be in the image at a specific point with a specific layout and to a specific scale. Chamfer distance matching is a favorable technique identifying image shapes. It is still extensively applied in pedestrian detection together with shape recognition. Detection of pedestrians can be viewed as a critical situation of human event detection, situations where people walk in upright positions (Ke, Sukthankar and Hebert).
These are shown in table 1 below:
Motion detection techniques comparison
Technique Features
Shape-matching Has spatio-temporal feature that is used to motion history images
Background subtraction may result in quality spatio images
Flow matching It acts on the spatio-temporal sequence in a direct way
Videos and templates can be correlated so that distant images can be captured
Space-time interest points Represents the general environment surrounding the local region.
Pose tracking relies on the video frame-by-frame process tracking and object segment from background clutter
3D chamfer matching places different same patient scans automatically in same co-ordinate system emanating from a specific part of the anatomy
can be applied in analyzing organ motion, tumor control follow up and verifying treatment delivery by use of electronic portal imaging and CT scans
CHAPTER 2METHODOLOGYINTRODUCTION
This chapter explains the existing situation awareness techniques, their efficiency and use. The situational awareness techniques covered in the chapter includes freeze probe, self-rating Real-time probe, and observer rating techniques. Several techniques of motion detection have also been analyzed, including optical flow and the Bayesian illumination-invariant motion detection
2.1 SITUATIONAL AWARENESS TECHNIQUES2.1.1 Situational Awareness requirements analysis This technique represents the first step in any assessment of SA. It is usually conducted in order to determine what actually determines operator SA in the task or environment under analysis. A generic procedure for conducting an SA requirements analysis that involves the use of unstructured interviews with SME’s (subject matter experts), goal-directed task analysis and questionnaires is described in order to determine the relevant SA requirements. The output of an SA requirements analysis is then used during the development of the SA assessment technique used, in order to determine which elements comprise operator SA and thus, what should be assessed.
2.1.2 Freeze probe techniques Freeze probe techniques entail the administration of Situation Awareness related queries on-line during ‘freezes’ in a simulation of the task under analysis. Classically, a task is randomly frozen and a set of Situation Awareness queries regarding the current situation are administered. The technique requires a participant to answer each query based upon his understanding of the situation, as per the freeze point. Throughout these ‘freezes’ all operator displays and windows are typically blanked. A computer system is used to choose and manage the queries and also to record the responses. One of the advantages related with the use of these techniques is their straightforward nature. Nevertheless, freeze probe techniques are criticized for their interference upon primary task performance, and its applications only where there is a simulation of the task below analysis.
Situation awareness global assessment technique (SAGAT); It is the most common freeze probe technique that was developed to assess pilot Situation Awareness across the three levels of SA anticipated in the three-level model. This technique is composed of a set of queries designed to assess participant SA, including level 1 SA (perception of the elements), level 2 SA) and level 3 SA (projection of future status). Even though SAGAT was developed particularly for use in the military aviation domain, a number of different versions of it exist, including a specific air-to-air tactical aircraft version, an advanced bomber aircraft version and an Air traffic control version SALSA; is another freeze probe technique that was developed specially for use in air traffic control. The SALSA queries are based upon fifteen aspects of aircraft flight, such as vertical tendency, flight level, ground speed, heading, conflict and type of conflict. SACRI which stands for Situation Awareness Control Room Inventory is a version of SAGAT and uses the freeze technique to manage control room based Situation Awareness queries.
2.1.3 Real-time probe techniquesThese techniques are alternative approach to the use of extremely intrusive freeze probe techniques. Real-time probe techniques involves the administration of Self Answer related queries on-line during the relevanat task performance, but without a freeze of the task under analysis. Usually, subject matter experts (SME’s) develop queries either before or during task performance and administer them, with no freeze, at the pertinent points throughout task performance.
The situation present assessment method (SPAM); is a real-time , probing technique that was developed for use in measuring of air traffic controllers SA. The technique involves the use of on-line real time probes to assess operator SA. The analyst probes the operator for SA using task associated SA queries based on relevant information in the environment through telephone. The response time for query, for correct responses ,shall be taken as an indicator of the operators Situation Awareness. Moreover, the time it takes to respond to the telephone call can acts as an indicator of workload. SASHA is a method developed by Euro control for the measurement of air traffic controllers SA in computerized systems. The methodology comprises of two techniques; SASHA_Q (post-trial questionnaire) and SASHA_L (real-time probe technique).
SASHA_L; is based upon the SPAM technique and involves questioning the participant on-line using real-time SA related queries. The response information and time taken for response is recorded. When the assessment is completed, the participant finishes the SASHA_Q questionnaire, which is composed of ten questions designed to gauge participant SA.
2.1.4 Self-rating techniquesThese techniques are used to increase a particular assessment of participant SA. They are usually administered after trial; self-rating techniques involve participants giving a subjective score of their perceived SA through an SA related rating scale. Some primary advantages of self-rating techniques are their simplicity of application (i.e. easy, quick and low cost) and their non-interfering nature (as they are administered after trial). Nevertheless, subjective self-rating techniques are heavily criticized for a number of reasons, which includes the various problems related with the collection of SA data post-trial and also issues concerning their sensitivity.
The situation awareness rating technique (SART); is a subjective rating technique that was initially developed for the evaluation of pilot SA.
The following are ten dimensions used by SART to measure an operator SA:
Familiarity with the situation, variability of the situation
information quality
focusing of attention
spare mental capacity,
information quantity,
instability of the situation,
concentration of attention,
complexity of the situation, and
arousal.
SART is administered after trial and it involves the participant rating each measurement on a seven point rating scale (1 = Low, 7 = High) so as to gain a particular measure of SA. The above ten SART dimensions can also be reduced into the quicker three dimensional SART, which involves participants rating
attention supply
attention demand
understanding.
Situation awareness rating scales technique (SARS); is also subjective rating technique that was designed for the military aviation domain. When using this technique (SARS), participants subjectively rate their performance on six-point rating scale for 31 facets of fighter pilot SA. The SARS SA categories and related behaviors were developed from interviews with experienced F-15 pilots. The thirty one SARS behaviors are divided into eight categories representing phases of mission performance These 8 categories are: tactical game plan (e.g. developing and executing plan), communication (e.g. quality), information interpretation (e.g. threat prioritization), tactical employment visual (e.g. threat evaluation), general traits (example: spatial ability, decisiveness), tactical employment beyond visual range (e.g. targeting decisions), and tactical employment general (e.g. defensive action).
The 31 SARS behaviors represent those that are essential to mission success. The Crew awareness rating scale (CARS) technique has been used to evaluate command and control commanders SA and workload. The CARS technique is made up of two separate sets of questions based upon the 3 level model of situation awareness.
The content subscale is made up of three statements designed to obtain ratings based upon ease of understanding, identification and projection of task SA elements. The fourth statement is designed to evaluate how well the participant identifies pertinent task associated goals in the situation. The workload subscale also comprises of four statements, which are designed to assess how difficult, it is for the participant in question to identify, in terms of mental effort, project the future states of the SA related elements in the situation. CARS are administered after trial and involve participants rating each category on a scale of one (ideal) to four (worst).
Mission Awareness Rating Scale (MARS) technique; is a development of the CARS technique designed particularly for use in the measurement of SA in military exercises. The MARS technique was developed for use in real world field settings, rather than in simulations of military training.
2.1.5 Observer rating techniques These techniques are commonly used to measure SA during tasks performed in-the-field. Observer rating techniques involves a subject matter expert (SME) observing participants performing the task under analysis and then provides an assessment of each participants SA. The ratings are based upon observable SA associated behavior exhibited by the participants during a particular task performance. The major advantages associated with the use of observer rating scales to assess SA are their non-intrusive features and their ability to be applied in-the-field. On the other hand, the extent to which observers can precisely rate participant SA is problematic, and also multiple subject matter experts (SME’s) are required.
Situation Awareness Behavioral Rating Scale (SABARS). Tis involves the observer rating technique , that was used to measure infantry personnel situation awareness in the field during training exercises.This proven technique requires the domain expert to observe participants during task performance in the field and rate them on 28 observable SA related behaviors. A five point rating scale (i.e. 1=Very poor, 5 =Very good) and other ‘not applicable’ category are used.
2.2 PERFORMANCE MEASURES These involve measuring relevant aspects of performance of a participant during the task under analysis. Basing upon the
