Embedded Files
MEMBERS INVOLVED: Vijay N Gangapure, Susmit Nanda
CAUSAL VIDEO SEGMENTATION USING SUPERSEEDS AND GRAPH MATCHING
SUPERPIXEL BASED CAUSAL MULTISENSOR VIDEO FUSION
Video surveillance systems have become extremely important in recent times. It has been observed that information extracted from a single spectrum video is often insufficient in adverse conditions like low illumination, shadowing, smoke, dust, unstable background, and camouflage. In addition, real-time video processing systems need to have very fast execution time, and in such systems, future frames may not be available at the time of processing the current frame. In this work, we propose a superpixel based causal multisensor video fusion algorithm which can be conveniently applied for real-time surveillance tasks. Superpixel level spatial and temporal saliency models are proposed for the visible and infra-red (input) video pairs. Superpixel level fusion rules are designed to obtain the fused output. Comprehensive comparison with similar methods on publicly available datasets clearly suggest the supremacy of the proposed algorithm.
Causal video segmentation methods use only past video frames to achieve the final segmentation. In this work, a novel framework forsemantic segmentation of causal video using superseeds and graph matching is proposed. We first employ SLIC for the extraction of superpixels in a causal video frame. A set of superseeds is chosen from the superpixels in each frame using color and
texture based spatial affinity measure. Temporal coherence is ensured through propagation of labels of the superseeds across each pair of adjacent frames. A
graph matching procedure based on comparison of the eigenvalues of graph Laplacians is employed for label propagation. Watershed algorithm is applied finally
to label the remaining pixels to achieve final segmentation.
MEMBERS INVOLVED: Vijay N Gangapure, Susmit Nanda
VIDEO SUMMARIZATION FOR MULTIMEDIA ANALYSIS
The problem of video summarization deals with succinct representation of a video. Design of video storyboards has emerged as an important area of research in recent times. Proper design of a video storyboard enables an user to access any video in a friendly and meaningful way. In this work, we aim to develop automated methods for construction of video storyboards and Video skimming based on Computational geometry tools and various statistical techniques. For more details, please visit the RESULTS link...
PRESENT MEMBERS: Sanjay K. Kuanar, Kunal Bhusan Ranga RESULTS
PAST MEMBERS: Rameswar Panda, Moloy Narayan Das
THEORETICAL VARIATIONS OF THE KNN ALGORITHM
Appropriate measures of distance and similarity are two prime issues in the field of pattern recognition. In the present work, we aim to address the above critical issues to improve the performance of the kNN algorithm and finding optimal value of k for kNN algorithm.
MEMBER INVOLVED: Gautam Bhattacharya
CO-ADVISOR: Dr. Koushik Ghosh
BLOOD VESSEL SEGMENTATION IN ZEBRA FISH IMAGES
The goal of this project is to perform morphometric analysis of pancreatic blood vessels in Zebra Fish images. This study will enable us to have a better understanding of the functioning of pancreas.
MEMBERS INVOLVED: Sudipto Mukherjee, Sarthak Chatterjee
COLLABORATORS: Dr. Enrico Grisan, Dr. Natascia Tiso
STEERABLE LOCAL FREQUENCY BASED MULTI-SPECTRAL MULTI-FOCUS IMAGE FUSION
We propose steerable local frequency map which can be used as focus measure in multispectral multifocus image fusion. The performance of the proposed focus measure is uniform and efficient across different spectra as compared to state-of-the-art focus measures. Further, based on the proposed focus measure, we propose multifocus image fusion algorithm . The comprehensive experimentation and results clearly demonstrates the promising results across visible, infra-red and thermal spectra.
MEMBERS INVOLVED: Vijay N Gangapure, Sudipta Bandyopadhyay
MUSCLE STEM CELL TRACKING USING PARTICLE FILTERING
Muscle satellite cells are the dedicated stem cells responsible for postnatal skeletal muscle growth, repair, and hypertrophy. In this work, our goal is to develop a semi-automated approach for satellite cell tracking. We are currently exploring Gaussian Mixture Models (GMM) for cell segmentation and Reversible Jump Markov Chain Monte Carlo (RJMCMC) methods for robust cell tracking.
MEMBER INVOLVED: Angshuman Paul
COLLABORATOR: Dr. Kannappan Palaniappan
SUBJECT INVARIANT GAIT RECOGNITION FROM SILHOUETTES
Gait is a behavioral biometric that measures the way people walk. The strength of gait, compared to other biometrics, is that it does not require cooperative subjects. In this work, we are focusing on development of covariate condition independent gait recognition techniques using the concept of pattern recognition and template matching.
MEMBER INVOLVED: Somnath Das
PELVIC FRACTURE DETECTION USING GRAPH CUTS AND CURVATURES
Traumatic injury of the pelvis is common and potentially devastating, with pelvic fractures being a major cause of trauma patient mortality. Detection of pelvic injuries is challenging due to varying injury patterns and resulting complications such as hemorrhage and infection. We are developing computer-aided detection (CAD) techniques for pelvic fracture detection.
We are now developing a model of the pelvic bone which can be beneficial in fracture detection and segmentation purposes.
PRESENT MEMBERS: Satyajit Neogi, Pulak Gandhi
PAST MEMBERS: Arka Mukherjee, Bhaskar Sen
COLLABORATORS: Dr. Joseph Burns, Dr. Ronald Summers, Dr. Jianhua Yao
NON-RIGID IMAGE REGISTRATION USING GRAPH CUTS
MEMBERS INVOLVED: Soumyadip Sengupta, Udit Halder
PEDESTRIAN TRACKING USING LEVEL SETS
In this work, we aim to explore tracking of a dynamic implicit interface by applying the Chan-Vese model within a narrow-band. Most importantly, we are developing temporal information in terms of variations in intensity and region enclosed by the target over successive video frames in the spatial level set framework.
Problem of non-rigid registration has become very important in the area of biomedical imaging. In this work, we aim to improve the graph cuts-based solution to non-rigid registration with a novel data term. This novel data term can efficiently handle the dissimilarities in the intensity patterns between the floating and the reference images which may also arise due to some changes in illumination in addition to motion. We are now exploring how non-rigid geometry can be added in the energy function of the graph cut to further improve the registration accuracy and computational time.
PRESENT MEMBERS: Debabrota Basu, Sayan Basu Roy
PAST MEMBERS: Suman Bose, Raunak Roy
COLLABORATOR: Dr. Ayman El-Baz
TRACKING OF HUMAN MONOCYTES USING MATCHING AND LINKING OF BIPARTITE GRAPHS
Automated visual tracking of cells from video microscopy has many important biomedical applications. In this work, we model the problem of cell tracking over pairs of video microscopy image frames as a minimum weight matching problem in bipartite graphs. The bipartite matching essentially establishes one-to-one correspondences between the cells in different frames.
MEMBERS INVOLVED: Rohit Chatterjee, Mayukh Ghosh
COLLABORATOR: Dr. Nilanjan Ray
MEDICAL IMAGE SEGMENTATION
CEREBRAL WHITE MATTER SEGMENTATION FROM LOW CONTRAST MRI USING MODIFIED GRAPH CUTS
Study of cerebral white matter in the brain is an important medical problem which helps in better understanding of brain disorders like autism. The goal of this research is to segment the cerebral white matter from the input Magnetic Resonance Imaging (MRI) data using a novel probabilistic graph cut algorithm.
MEMBERS INVOLVED: Ashish K. Rudra, Mainak Sen
COLLABORATOR: Dr. Ayman El-Baz
KIDNEY SEGMENTATION FROM LOW CONTRAST MRI USING GRAPH CUTS AND PIXEL CONNECTIVITY
Kidney segmentation from abdominal MRI data is used as an effective and accurate indicator for renal function in many clinical situations. The goal of this research is to accurately segment kidney from very low contrast MRI data. The present problem becomes challenging problem mainly due to poor contrast, high noise and partial volume effects introduced during the scanning process. problem mainly due to poor contrast, high noise and partial volume effects introduced during the scanning process. In this work, we propose a novel segmentation algorithm using graph cuts and pixel connectivity. MEMBER INVOLVED: Ashish K. Rudra
COLLABORATOR: Dr. Ayman El-Baz
DETECTION OF RETINAL BLOOD VESSELS
In this work our target is to extract blood vessel in retinal images by using eigenvalue analysis of the Hessian matrix and Matched Filtering. A vesselness measure is obtained from the eigenvalues of the Hessian matrix. Matched filtering gives us the the maximum response of the favaourable pixel. We have exploited the synergism between these two very different approaches using orientation histogram.
Currently, we are focusing on detection of micro-aneurysms in these retinal blood vessels.
PRESENT MEMBERS: Satrajit Mukherjee, Kunal Pal
PAST MEMBERS: Dhiraj K. Jha, Tapabrata Chakraborti
COLLABORATOR: Dr. Xiaoyi Jiang
Page updated
Google Sites
Report abuse