Kan Hong

Detection and classification of stress using thermal imaging technique

This paper reports how Electro-Optics (EO) technologies such as thermal and hyperspectral [1-3] imaging methods can be used for the detection of stress remotely. Emotional or physical stresses induce a surge of adrenaline in the blood stream under the command of the sympathetic nerve system, which, cannot be suppressed by training. The onset of this alleviated level of adrenaline triggers a number of physiological chain reactions in the body, such as dilation of pupil and an increased feed of blood to muscles etc. The capture of physiological responses, specifically the increase of blood volume to pupil, have been reported by Pavlidiss pioneer thermal imaging work [4-7] who has shown a remarkable increase of skin temperature in the periorbital region at the onset of stress. Our data has shown that other areas such as the forehead, neck and cheek also exhibit alleviated skin temperatures dependent on the types of stressors. Our result has also observed very similar thermal patterns due to physical exercising, to the one that induced by other physical stressors, apparently in contradiction to Pavlidiss work [8]. Furthermore, we have found patches of alleviated temperature regions in the forehead forming patterns characteristic to the types of stressors, dependent on whether they are physical or emotional in origin. These stress induced thermal patterns have been seen to be quite distinct to the one resulting from having high fever.

Target recognitions in multiple-camera closed-circuit television using color constancy

Abstract. People tracking in crowded scenes from closed-circuit television (CCTV) footage has been a popular and challenging task in computer vision. Due to the limited spatial resolution in the CCTV footage, the color of people’s dress may offer an alternative feature for their recognition and tracking. However, there are many factors, such as variable illumination conditions, viewing angles, and camera calibration, that may induce illusive modification of intrinsic color signatures of the target. Our objective is to recognize and track targets in multiple camera views using color as the detection feature, and to understand if a color constancy (CC) approach may help to reduce these color illusions due to illumination and camera artifacts and thereby improve target recognition performance. We have tested a number of CC algorithms using various color descriptors to assess the efficiency of target recognition from a real multicamera Imagery Library for Intelligent Detection Systems (i-LIDS) data set. Various classifiers have been used for target detection, and the figure of merit to assess the efficiency of target recognition is achieved through the area under the receiver operating characteristics (AUROC). We have proposed two modifications of luminance-based CC algorithms: one with a color transfer mechanism and the other using a pixel-wise sigmoid function for an adaptive dynamic range compression, a method termed enhanced luminance reflectance CC (ELRCC). We found that both algorithms improve the efficiency of target recognitions substantially better than that of the raw data without CC treatment, and in some cases the ELRCC improves target tracking by over 100% within the AUROC assessment metric. The performance of the ELRCC has been assessed over 10 selected targets from three different camera views of the i-LIDS footage, and the averaged target recognition efficiency over all these targets is found to be improved by about 54% in AUROC after the data are processed by the proposed ELRCC algorithm. This amount of improvement represents a reduction of probability of false alarm by about a factor of 5 at the probability of detection of 0.5. Our study concerns mainly the detection of colored targets; and issues for the recognition of white or gray targets will be addressed in a forthcoming study.

Illumination invariance and shadow compensation via spectro-polarimetry technique

Abstract. A major problem for obtaining target reflectance via hyperspectral imaging systems is the presence of illumination and shadow effects. These factors are common artefacts, especially when dealing with a hyperspectral imaging system that has sensors in the visible to near infrared region. This region is known to have highly scattered and diffuse radiance that can modify the energy recorded by the imaging system. A shadow effect will lower the target reflectance values due to the small radiant energy impinging on the target surface. Combined with illumination artefacts, such as diffuse scattering from the surrounding targets, background or environment, the shape of the shadowed target reflectance will be altered. We propose a new method to compensate for illumination and shadow effects on hyperspectral imageries by using a polarization technique. This technique, called spectro-polarimetry, estimates the direct and diffuse irradiance based on two images taken with and without a polarizer. The method is then evaluated using a spectral similarity measure, angle and distance metric. The results of indoor and outdoor tests have shown that using the spectro-polarimetry technique can improve the spectral constancy between shadow and full illumination spectra.

Remote sensing of stress using electro-optics imaging technique

Emotional or physical stresses induce a surge of adrenaline in the blood stream under the command of the sympathetic nerve system, which, cannot be suppressed by training. The onset of this alleviated level of adrenaline triggers a number of physiological chain reactions in the body, such as dilation of pupil and an increased feed of blood to muscles etc. This paper reports for the first time how Electro-Optics (EO) technologies such as hyperspectral [1,2] and thermal imaging[3] methods can be used for the detection of stress remotely. Preliminary result using hyperspectral imaging technique has shown a positive identification of stress through an elevation of haemoglobin oxygenation saturation level in the facial region, and the effect is seen more prominently for the physical stressor than the emotional one. However, all results presented so far in this work have been interpreted together with the base line information as the reference point, and that really has limited the overall usefulness of the developing technology. The present result has highlighted this drawback and it prompts for the need of a quantitative assessment of the oxygenation saturation and to correlate it directly with the stress level as the top priority of the next stage of research.

Enhanced Object Recognition in Cortex-Like Machine Vision

This paper reports an extension of the previous MIT and Caltech’s cortex-like machine vision models of Graph-Based Visual Saliency (GBVS) and Feature Hierarchy Library (FHLIB), to remedy some of the undesirable drawbacks in these early models which improve object recognition efficiency. Enhancements in three areas, a) extraction of features from the most salient region of interest (ROI) and their rearrangement in a ranked manner, rather than random extraction over the whole image as in the previous models, b) exploitation of larger patches in the C1 and S2 layers to improve spatial resolutions, c) a more versatile template matching mechanism without the need of ‘pre-storing’ physical locations of features as in previous models, have been the main contributions of the present work. The improved model is validated using 3 different types of datasets which shows an average of ~7% better recognition accuracy over the original FHLIB model.

A biological cortex like target recognition and tracking in cluttered background

This paper reports how objects in street scenes, such as pedestrians and cars, can be spotted, recognised and then subsequently tracked in cluttered background using a cortex like vision approach. Unlike the conventional pixel based machine vision, tracking is achieved by recognition of the target implemented in neuromorphic ways. In this preliminary study the region of interest (ROI) of the image is spotted according to the salience and relevance of the scene and subsequently target recognition and tracking of the object in the ROI have been performed using a mixture of feed forward cortex like neuromorphic algorithms together with statistical classifier & tracker. Object recognitions for four categories (bike, people, car & background) using only one set of ventral visual like features have achieved a max of ~70% accuracy and the present system is quite effective for tracking prominent objects relatively independent of background types. The extension of the present achievement to improve the recognition accuracy as well as the identification of occluded objects from a crowd formulates the next stage of work.

Colour invariant target recognition in multiple camera CCTV surveillance

People tracking in crowded scene have been a popular, and at the same time a very difficult topic in computer vision. It is mainly because of the difficulty for the acquisition of intrinsic signatures of targets from a single view of the scene. Many factors, such as variable illumination conditions and viewing angles, will induce illusive modification of intrinsic signatures of targets. The objective of this paper is to verify if colour constancy (CC) approach really helps people tracking in CCTV network system. We have testified a number of CC algorithms together with various colour descriptors, to assess the efficiencies of people recognitions from multi-camera i-LIDS data set via receiver operation characteristics (ROC). It is found that when CC is applied together with some form of colour restoration mechanisms such as colour transfer, it does improve people recognition by at least a factor of 2. An elementary luminance based CC coupled with a pixel based colour transfer algorithm have been developed and it is reported in this paper.