Vision, Image  & Signal Processing

Research Group @ NIT Kurukshetra India

Visual Surveillance

International usage and interest in surveillance of public spaces is growing at an unprecedented pace in response to global terrorism. The research being conducted by the Security and Surveillance Group seeks to overcome the inadequacies of current surveillance techniques and meet the increased demand for improvements in human monitoring, following the immense costs of successful terrorist attacks. A solution may be found in advanced surveillance systems employing computer monitoring of all video feeds, delivering the alerts to human responders for triage. Indeed such systems may assist in maintaining the high level of vigilance required over many years to detect the rare events associated with terrorism - a well designed computer system is never caught off guard.

Our surveillance research is aimed at developing advanced security solutions based on computer vision and pattern recognition technologies and advanced embedded systems. These systems have been evaluated at several sites around Australia and overseas. This research has attracted national and international interest from agencies and has secured over $1.6 million in funding.

In addition to this theme is continued research into robust biometrics suitable for CCTV and mobile deployment.  The team has won a number of awards for its biometrics research and is well-embedded in the international research community.  Our technologies extend to medical surveillance via our Digital Pathology Project with Sullivan and Nicolaides Pathology as well as ionospheric surveillance via our projects on Space Weather with AOARD.

As a leading computer vision and pattern recognition research team our targetted conferences for our fundamental research are CVPR, ICCV, and ECCV and the journals PR, TPAMI, and TIP.  We welcome applications from strong PhD Candidates who wish to pursue top level academic research and excellence.

Image Processing

Extracting information from a digital image often depends on first identifying desired objects or breaking down the image into homogenous regions (a process called 'segmentation') and then assigning these objects to particular classes (a process called 'classification'). This is a fundamental part of computer vision, combining image processing and pattern recognition techniques. Homogeneous may refer to the color of the object or region, but it also may use other features such as texture and shape. The methodology can be used to identify tumours in medical images, crops in satellite imagery, cells in biological tissue, or human faces in standard digital images or video. Each segmentation/classification implementation has the same fundamental approach; however, specific objects and imagery often require dedicated techniques for improved success. In the VIP lab, a dedicated example of segmentation is our advanced work in decoupled active contours. A dedicated example of classification is the automated identification of sea ice in satellite SAR images.

Affective Computing

The new on-going mega-trend of ubiquitous computing has resulted in that the research on human-centred computing systems has become a major topic of interest both in academia and industry. This means that the future technological solutions for human-computer interaction (HCI) must have better capabilities for understanding human behaviour than is currently the case. The need has created a new field of research, affective computing, aiming to improve the HCI by including the interpretation of the human emotions to the functionality of machines. By adapting to the user’s emotional state, more appropriate responses could be given.

In affective computing one of the main goals is the automatic recognition of human emotions. Emotions are fundamental for humans, impacting everyday activities such as perception, communication, learning and decision-making. In addition to speech, they are expressed through gestures, facial expressions and other non-verbal clues. Many physiological signals also contain information about the emotional state of humans. Various sensor signals can therefore be used when developing automatic emotion recognition capabilities for computers.

The main objective of the project is to develop and combine new methods in machine vision and biosignal processing to detect and classify the emotional state of humans. The proposed multimodal approach for automatic emotion recognition is unique in Finland and very rare globally. The project aims to create a technology applicable to measure emotional responses to advertisements and medical diagnostics of affective disorders.