Anirban Dasgupta

A Vision-Based System for Monitoring the Loss of Attention in Automotive Drivers

Onboard monitoring of the alertness level of an automotive driver has been challenging to research in transportation safety and management. In this paper, we propose a robust real-time embedded platform to monitor the loss of attention of the driver during day and night driving conditions. The percentage of eye closure has been used to indicate the alertness level. In this approach, the face is detected using Haar-like features and is tracked using a Kalman filter. The eyes are detected using principal component analysis during daytime and using the block local-binary-pattern features during nighttime. Finally, the eye state is classified as open or closed using support vector machines. In-plane and off-plane rotations of the drivers face have been compensated using affine transformation and perspective transformation, respectively. Compensation in illumination variation is carried out using bihistogram equalization. The algorithm has been cross-validated using brain signals and, finally, has been implemented on a single-board computer that has an Intel Atom processor with a 1.66-GHz clock, a random access memory of 1 GB, ×86 architecture, and a Windows-embedded XP operating system. The system is found to be robust under actual driving conditions.

Quantified Computation Tree Logic

Computation Tree Logic (CTL) is one of the most syntactically elegant and computationally attractive temporal logics for branching time model checking. In this paper, we observe that while CTL can be verified in time polynomial in the size of the state space times the length of the formula, there is a large set of reachability properties which cannot be expressed in CTL, but can still be verified in polynomial time. We present a powerful extension of CTL with first-order quantification over sets of reachable states. The extended logic, QCTL, preserves the syntactic elegance of CTL while enhancing its expressive power significantly. We show that QCTL model checking is PSPACE-complete in general, but has a rich fragment (containing CTL) which can be checked in polynomial time. We show that this fragment is significantly more expressive than CTL while preserving the syntactic beauty of CTL.

Video & EOG based investigation of pure saccades in human subjects

Human Computer Interaction (HCI) is the methodology through which computing systems understand actions of human beings. Some important applications of HCI as reported in literature are eye gaze estimation, eye tracking, alertness and health monitoring etc. Of them major applications are based upon eye movements. However, accurate estimation of these measurements is still a challenge in research community. A particular type of eye movement i.e. saccadic eye movement has potential applications in HCI such as alertness assessment, disease diagnosis etc. Electrooculography (EOG) based measurement of saccadic movements has been reported to be an accurate method. However being a contact based method, alternative method as suggested by literature is image based approach. This paper aims to investigate the correlation between video and EOG based observation of pure horizontal directional saccades to establish a relationship between image and EOG signals.

A two-stage framework for denoising electrooculography signals

Abstract Denoising of electrooculography (EOG) signals is a challenging task as the noise and signal share the same frequency band. This paper proposes a two-stage framework for denoising EOG signals. The first stage approach is based on preserving the nature of eye movements while the second stage is based on the nature of noise (Gaussian or not). In the first stage, denoising is carried out using one out of four filtering methods, each filter being optimal for a particular EOG pattern. The four methods used in the first stage are linear bandpass filtering, stationary wavelet transform (SWT), empirical mode decomposition (EMD) and median filtering. The Stage I framework selects the output that provides the highest estimated signal to noise ratio (SNR). In case, the Stage I filtering does not provide a significant SNR, the system uses Stage II filtering. In the second stage, we use two recursive state estimators, i.e. a Kalman filter and a particle filter for further denoising. The two-stage method is found to provide a better SNR as compared to a single stage method.

A real time algorithm for detection of spectacles leading to eye detection

Eye detection plays an important role in many Intelligent Human Computer Interaction (IHCI) applications such as eye tracking, eye gaze tracking, drowsiness monitoring etc. However the presence of spectacles makes the eye detection task difficult because of the glint produced by the lens. Earlier approaches on detection of eyes occluded with spectacles have been reported in literature, however with limited accuracies. This paper proposes a real-time algorithm for detection of spectacles leading to the detection of eyes. A video database of human subjects with and without spectacles has been created to evaluate the performance of the algorithm.

A video database of human faces under near Infra-Red illumination for human computer interaction applications

Human Computer Interaction (HCI) is an evolving area of research for coherent communication between computers and human beings. Some of the important applications of HCI as reported in literature are face detection, face pose estimation, face tracking and eye gaze estimation. Development of algorithms for these applications is an active field of research. However, availability of standard database to validate such algorithms is insufficient. This paper discusses the creation of such a database created under Near Infra-Red (NIR) illumination. NIR illumination has gained its popularity for night mode applications since prolonged exposure to Infra-Red (IR) lighting may lead to many health issues. The database contains NIR videos of 60 subjects in different head orientations and with different facial expressions, facial occlusions and illumination variation. This new database can be a very valuable resource for development and evaluation of algorithms on face detection, eye detection, head tracking, eye gaze tracking etc. in NIR lighting.

Analysis of training parameters for classifiers based on Haar-like features to detect human faces

This paper analyzes the performance of the Haar-like feature based classifier for detection of face with fewer features. The lower dimensional feature space representation of the image may reduce the computational burden compromising the accuracy in detection of faces with varying orientations. In this work we train the classifier with positive instances of different orientations under such feature constraint. The training parameters like maximum deviation and maximum angle are varied to form different classifiers. Experimental results show optimum values of the design parameters can produce good performance of the classifier to detect frontal as well as tilted human faces.

Evaluation of denoising techniques for EOG signals based on SNR estimation

This paper evaluates four algorithms for denoising raw Electrooculography (EOG) data based on the Signal to Noise Ratio (SNR). The SNR is computed using the eigenvalue method. The filtering algorithms are a) Finite Impulse Response (FIR) bandpass filters, b) Stationary Wavelet Transform, c) Empirical Mode Decomposition (EMD) d) FIR Median Hybrid Filters. An EOG dataset has been prepared where the subject is asked to perform letter cancelation test on 20 subjects.

Identification of eye saccadic signatures in electrooculography data using time-series motifs

This paper proposes a method for detecting eye saccadic signatures from Electrooculograms (EOG). Saccadic movements bear a particular pattern in the raw EOG time-series data. In this work, we interpret this signature of saccades as time-series motifs. The major issue in finding a saccade using standard motif identification methods is that saccadic signatures may be shrunk or stretched in the EOG time-series based on the saccadic duration. This shrinkage or stretching issue has been overcome in this work by using Dynamic Time Warping (DTW). This approach has been tested on our created EOG database and compared with existing methods of saccade detection in EOG. The proposed method shows higher accuracy as compared to existing ones. However, the execution time poses to be a limitation of the proposed method, which may be overcome by using high-speed processors and parallel computing platforms.

A Multimodal System for Assessing Alertness Levels Due to Cognitive Loading

This paper proposes a scheme for assessing the alertness levels of an individual using simultaneous acquisition of multimodal physiological signals and fusing the information into a single metric for quantification of alertness. The system takes electroencephalogram, high-speed image sequence, and speech data as inputs. Certain parameters are computed from each of these measures as indicators of alertness and a metric is proposed using a fusion of the parameters for indicating alertness level of an individual at an instant. The scheme has been validated experimentally using standard neuropsychological tests, such as the Visual Response Test (VRT), Auditory Response Test (ART), a Letter Counting (LC) task, and the Stroop Test. The tests are used both as cognitive tasks to induce mental fatigue as well as tools to gauge the present degree of alertness of the subject. Correlation between the measures has been studied and the experimental variables have been statistically analyzed using measures such as multivariate linear regression and analysis of variance. Correspondence of trends obtained from biomarkers and neuropsychological measures validate the usability of the proposed metric.