Image-based Plant Phenotyping Analysis: Image based plant phenotyping has the following desirable characteristics relative to many other methods of collecting plant phenotypic data: 1) non-invasive, i.e., phenotypic traits can be measured without damaging the plants being measured; 2) tractable to automation, i.e., with properly engineered systems, very little, or no manual intervention or physical human labor is required; 3) scalable, i.e., large population of plants can be analyzed in short period of time. However, extracting meaningful numerical phenotypes from plant images remains a substantial bottleneck in image-based automated plant phenotyping.
Our university is equipped with Lemnatec Scanalyzer 3D high throughput plant phenotyping system which is used to capture multi-sensor phenotyping measurements of plants in a non-destructive fashion on a daily basis. Our conveyor system has the capacity to host 672 plants with heights up to 2.5 meters. It has three watering stations with balance can add water to target weight or specific volume and records the specific quantity of water added on a daily basis. Images are captured by four different types of cameras,(a) visible light side view and top view, (b) infrared side view and top view, (c) fluorescent side view and top view, and (d) hyperspectral side view and near infrared top view. My research includes computation of plant phenotypes from the multi-modal image sequences.
Biometrics based human identification: The necessity of human identification has increased dramatically due to growing importance of adequate security measures for well-fare of the mankind. In response to this necessity, biometrics has emerged as a reliable means of identifying a human subject based on the distinguishable biological features. Physiological biometrics analyses the physical attributes of a subject such as face, fingerprint, iris, palm-print, earlobe geometry etc., and require cooperation from the subject for a particular view in a controlled lighting environment. Behavioral biometrics, e.g., gait examines human action performed over a defined time interval for identification.
Unlike other physiological biometrics, face data can be captured without requiring the subject to physically touch the sensor, and hence only physiological biometrics to be used for remote human identification.
Medical image processing: Lung cancer is mainly caused by tobacco smoking and is a life-threatening disease for men and women all over the world. According to the survey held in 2012, 63759 people out of 70275 people in India who were affected by lung cancer, suffered death.
Thus, an automatic risk factor prediction model of lung cancer from lung nodules plays a very important role in the successful treatment of this disease. X-ray computed tomography (CT) has been widely adopted as a tool for lung cancer diagnosis. Although, CT images enable early detection of nodules efficiently, it is much more expensive compared to conventional chest X-ray images. Thus the proposed project, aims to develop a software-based automated cancer risk prediction model of the lung nodule by analyzing low-cost digital x-ray images using a retrospective study.
The project will be of enormous help especially in the rural areas with limited medical facilities. In the case of unavailability of expert oncologists in the rural areas, the project will be helpful for early prediction of lung cancer risk, so that the patient with risky cancer nodule will be quickly transferred to the hospital for proper follow-up treatment. In this way, the proposed project will make significant contributions towards the early detection of chest cancer from the low-cost x-ray images to increase the survival rate of cancer patients.
The art of information hiding-steganography: The word steganography is derived from two Greek roots-‘steganos’ ,meaning ‘covered’ , and ‘grapheion’ meaning ‘to write’. Steganography is the art of concealing the presence of information within an innocuous container (image, audio and video). I would like to conduct projects on innovative techniques of steganography for information hiding in order to make contributions in the field of network security.