Research and Development (2011-2016)

Theories of approximation of Euclidean metrics by digital distances from the properties of their hyperspheres.  

A novel approach exploiting properties of hyperspheres of digital distances in arbitrary dimension has been adopted for approximating Eulcidean metrics. New classes of distance functions called weighted t-cost distances (WtD), and linear combination of weighted t-cost and chamfering weighted distances (WtCWD) have been proposed and expressions for computing their hyperspheres in arbitrary dimensions have been derived. New results on computation of vertices, hypervolumes and hypersurfaces of hyperspheres conventional weighted or chamfering distance functions in arbitrary dimensions are also derived. It is shown how an analytical error measure, called maximum relative error (MRE) with respect to Euclidean metric, can be computed from these hyperspheres. The theory is applied to obtain very close approximators of Euclidean metric in arbitrary dimensions (MRE values ranging from approx. 1% to 6% for dimensions 2 to 100). It is shown that a class of generalized octagonal distances could be modeled by WtDs with a very low approximation error. Hence, the same analysis is also extended to this class of distances and MRE values for this class of distance functions also are computed to provide good approximators of Euclidean metric in arbitrary dimension.

Related Publications

  • J. Mukherjee (2011), On approximating Euclidean metrics by weighted t-cost distances in arbitrary dimension, Pattern Recognition Letters, 32, 824-831.
  • J. Mukherjee (2013), Hyperspheres of Weighted distances in arbitrary dimension, Pattern Recognition Letters, 34 (2), 117-123.
  • J. Mukherjee (2013), Linear combination of norms in improving approximation of Euclidean norm, Pattern Recognition Letters, 34(12), 1348-1355.
  • J. Mukherjee (2014), Linear combination of weighted t-cost and chamfering weighted distances, Pattern Recognition Letters, 40, pp. 72-79.
  • J. Mukherjee (2016), Error analysis of octagonal distances defined by periodic neighborhood sequences for approximating Euclidean metrics in arbitrary dimension, Pattern Recognition Letters, Pattern Recognition Letters 75, pp. 16-23.
 

Signal, image and video processing techniques for neonatal health care  

Different techniques and algorithms for processing signal, image and video for facilitating neonatal health care have been developed. These include analysis of neonatal EEG signals, development of image and video processing tools for neurological examinations of babies, video processing to detect sleep-apnea of neonates, etc.

Related Publications

  • Sourya Bhattacharyya, Jayanta Mukhopadhyay, Arun K. Majumdar, Bandana Majumdar, Arun Kumar Singh, Chanchal Saha: Automated Burst Detection in Neonatal EEG. BIOSIGNALS 2011: 15-21.
  • Debi P. Dogra, Karthik Nandam, Arun K. Majumdar, Shamik Sural, Jayanta Mukherjee, Suchandra Mukherjee and Arun Singh (2012), Toward Automating Hammersmith Pulled-To-Sit Examination of Infants using Feature Point based Video Object Tracking, IEEE Transactions on Neural Systems & Rehabilitation Engineering, Vol. 20, No. 1, Jan., pp. 38-47 .
  • Sourya Bhattacharyya, Arunava Biswas, Jayanta Mukherjee, Arun Kumar Majumdar, Bandana Majumdar, Suchandra Mukherjee, and Arun Kumar Singh (2011), Feature Selection for Automatic Burst Detection in Neonatal Electroencephalogram IEEE Journal on Emerging and Selected Topics in Circuits and Systems Vol. 1, No. 4, Dec., page: 469-479.
  • D.P. Dogra, A. K. Majumdar, S. Sural, J. Mukherjee, S. Mukherjee, and A. K. Singh (2012), Analysis of Adductors Angle Measurement in Hammersmith Infant Neurological Examinations using Mean Shift Segmentation and Feature Point based Object Tracking Authors, Computers in Biology and Medicine, Elsevier, 42 (9), 925-934.
  • D.P. Dogra, V. Badri, A. K. Majumdar, S. Sural, J. Mukherjee, S. Mukherjee, and A. K. Singh (2014), Video analysis of Hammersmith lateral tilting examination using Kalman filter guided multi-path tracking, Medical & Biological Engineering & Computing, Springer, 52(9): 759-772.
  • Shashank Sharma, Sourya Bhattacharyya, Jayanta Mukherjee, Parimal Kumar Purkait, Arunava Biswas, Alok Kanti Deb (2015): Automated detection of newborn sleep apnea using video monitoring system. ICAPR 2015: 1-6
 

Medical Image Processing  

A technique for anatomical model guided brain MR segmentation has been developed. In this technique, correspondences with cross-sections of brain MR volume have been established (aligned) with those of an anatomical model, and different segments of 3-D volume have been identified. An efficient algorithm for computing cross-sections from triangular mesh model of an object has also been developed. This is used in the alignment of cross-sections of brain MR image with those of a triangular mesh model of the brain. There are also ongoing collaborative work for analyzing cone beam CT (CBCT) images of patients suffering from lung or colo-rectal cancer and under radiotherapeutic regime. A technique for measuring uncertainties in the positioning of pelvic region of patients has been developed. An algorithm for registering CBCT volumes has also been developed, which also takes care of deformation in volumes, as observed in the images of lungs.

Related Publications

  • Prasenjit Mondal, Jayanta Mukhopadhyay, Shamik Sural, Pinak Pani Bhattacharyya: An efficient model-guided framework for alignment of brain MR image sequences. SMC 2012: 2201-2206.
  • Prasenjit Mondal, Jayanta Mukhopadhyay, Shamik Sural, Pinak Pani Bhattacharyya: Anatomical Model-Guided Segmentation of Brain MR Image Sequences. ICVGIP 2014: 8:1-8:8
  • P. Mondal, P. Bhowmick, J. Mukherjee, and Shamik Sural (2015) Efficient computation of cross-sections from human brain model by geometric processing, Journal of Real-Time Image Processing, DOI 10.1007/s11554-015-0495-5
  • S. Mandal, Bijju K.V., J. Mukherjee, P.P. Das and I. Mallick (2015), Study of Variation of Pelvis Positioning for Patients Suffering from Rectal Cancer using Daily Kilo- Voltage Cone Beam CT Images, National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG-2015), IIT Patna, 16-19 Dec.
  • Bijju K.V., J. Mukherjee, P.P. Das, S. Chatterjee, S.N. Ray, and P. Sen (2015), Multimodal Image Registration of Lung images, National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG-2015), IIT Patna, 16-19 Dec.
  • Sai Phani Kumar Malladi, Bijju Kranthi Veduruparthi, Jayanta Mukherjee, Partha Pratim Das, Saswat Chakrabarti, Indranil Mallick (2016): Reduction of variance of observations on pelvic structures in CBCT images using novel mean-shift and mutual information based image registration, ICVGIP 2016: 84:1-84:8
 

Gait analysis  

Various techniques for analyzing human gait have been developed. They include both fronto-parallel and frontal gait recognition using optical video camera and Microsoft Kinect depth sensing imaging system, respectively. Frontal gait recognition using RGBD camera is one of the early works in this area, and is filed for patenting. One of its targeted applications is the security check in airport terminals.

Related Publications

  • A. Roy, S. Sural, J. Mukherjee, and G. Rigoll (2011), Gait Silhouette Reconstruction from Occluded Scenes, Hugo Proen, Eliza Yingzi Du, Jacob Scharcanski (Eds.); Springer Signal Image and Video Processing, Special Issue On Unconstrained Biometrics: Advances and Trends, volume 5, number 3, September, ISSN: 1863-1703.
  • A. Roy, S. Sural, and J. Mukherjee (2012), Gait Recognition using Pose Kinematics and Pose Energy Image Signal Processing, Elsevier, 92(3), pp. 780-792.
  • A. Roy, S. Sural, and J. Mukherjee (2012), Hierarchical Method Combining Gait and Phase of Motion with Spatiotemporal Model for Person Reidentification, Pattern Recognition Letters, 33 (14), 1891-1901.
  • P. Chattopadhyay, A. Roy, S. Sural, and J. Mukhopadhyay (2014), Pose Depth Volume Extraction from RGB-D Streams for Frontal Gait Recognition, Journal of Visual Communication and Image Representation, 25 (1), 53-63.
  • P. Chattopadhyay, S. Sural, and J. Mukherjee (2014), Frontal Gait Recognition from Incomplete Sequences using RGB-D Camera, IEEE Transactions on Information Forensics & Security, 9(11):1843-1856.
  • A. Roy, P. Chattopadhyay, S. Sural, J. Mukherjee, G. Rigoll (2015) Modelling, synthesis and characterisation of occlusion in videos, IET Computer Vision, 9(6): 821-830, 2015.
  • P. Chattopadhyay, S. Sural, and J. Mukherjee (2015), Frontal Gait Recognition from Occluded Scenes, Pattern Recognition Letters, 63: 9-15.
  • P. Chattopadhyay, S. Sural, and J. Mukherjee (2015), Information fusion from multiple cameras for gait-based re-identification and recognition, IET Image Processing,9(11): 969-976.
  • Pratik Chattopadhyay, Shamik Sural, Jayanta Mukherjee (2014): Exploiting Pose Information for Gait Recognition from Depth Streams. ECCV Workshops (1): 341-355.
 

Local rank transform and Distributed Video Coding  

Local rank transforms (LRT) are proposed to be used for representing and processing images, such as color image enhancement, edge extraction, etc. An interesting convergence property of LRTs in the form of an attractor has been demonstrated and theoretically established in one dimension. The property in higher dimension is yet to be proved. LRT based representation is found to be useful in distributed video coding (DVC). Algorithms for reconstructing images from their LRT have been developed for this purpose.

Related Publications

  • J. Mukherjee (2011), Local Rank Transform: Properties and Applications, Pattern Recognition Letters, 32, 1001-1008.
  • Kallol Mallick, and Jayanta Mukherjee (2014): Distributed Video Coding using Local Rank Transform. ICVGIP 2014: 18:1-18:8
  • Pudi Raj Bhagath, Kallol Mallick, Jayanta Mukherjee, Sudipta Mukopadhayay (2017): Low-complexity feedback-channel-free distributed video coding using local rank transform. IET Image Processing 11(2): 126-134.
  • Pudi Raj Bhagath, Jayanta Mukherjee, Sudipta Mukopadhayay (2016): Low complexity encoder for feedback-channel-free distributed video coding using deep convolutional neural networks at the decoder. ICVGIP 2016: 44:1-44:7
 

Document Image Analysis  

Algorithm for performing various preprocessing tasks, such as, skew correction, lay out analysis, removal of margin noise, underline, stamps, etc., have been developed. In some of these algorithms digital geometry based computation has been adopted. A technique for document retargeting, a novel concept introduced in document image analysis, has been developed. In document retargeting, a scanned document of a given lay-out, is converted to a desired layout, convenient for reading it in different display devices (e.g. in mobile devices). A novel segmentation algorithm using hypothesis testing and consensus analysis has also been developed.

Related Publications

  • Sanjoy Pratihar, Partha Bhowmick, Shamik Sural, Jayanta Mukhopadhyay (2012): Detection and removal of hand-drawn underlines in a document image using approximate digital straightness. DAR@ICVGIP 2012: 124-131
  • Soumyadeep Dey, Jayanta Mukhopadhyay, Shamik Sural, Partha Bhowmick (2012): Margin noise removal from printed document images. DAR@ICVGIP 2012: 86-93.
  • S. Pratihar, P. Bhowmick, S. Sural, and J. Mukhopadhyay (2013), Skew correction of document images by rank analysis in Farey sequence, International Journal of Pattern Recognition and Artificial Intelligence 27 (7), 1353004 (1-35 pages).
  • Soumyadeep Dey, Jayanta Mukhopadhyay, Shamik Sural, Partha Bhowmick (2013): Re-targeting of multi-script document images for handheld devices. MOCR@ICDAR 2013: 10:1-10:5
  • Soumyadeep Dey, Jayanta Mukherjee, Shamik Sural, Partha Bhowmick: Colored Rubber Stamp Removal from Document Images. PReMI 2013: 545-550
  • Soumyadeep Dey, Jayanta Mukherjee, Shamik Sural (2016): Consensus-based clustering for document image segmentation. IJDAR 19(4): 351-368 (2016)
  • Soumyadeep Dey, Jayanta Mukhopadhyay, Shamik Sural (2016): Removal of Gray Rubber Stamps. DAS 2016: 210-214
  • Amit Vijay Nandedkar, Jayanta Mukhopadhyay, Shamik Sural (2015): Text-graphics separation to detect logo and stamp from color document images: A spectral approach. ICDAR 2015: 571-575
 

Assistive Technology for visually challenged persons  

Electronic smart canes and camera based wearable implements have been developed for aiding navigation of visually challenged persons. Usability trials of smart canes had been carried out with a number of blind subjects under the supervision of doctors in a medical college and they were found to be very effective. An international patent of this device has been granted.

Related Publications

  • A. Kumar, R. Patra, M. Mahadevappa, J. Mukhopadhyay, and A.K. Majumdar (2013), An embedded system for aiding navigation of visually impaired persons, Current Science, Vol. 104, No. 3, Feb. 10, 2013, 302-306.
  • S. Bhatlawande, A. Sunkari, M. Manjunatha, J. Mukhopadhyay, M. Biswas, D. Das, and S. Gupta (2014) Electronic Bracelet and Vision Enabled Waist-belt for Mobility of Visually Impaired People, Assistive Technology, Taylor & Francis, 26(4), 186-195.
  • S. Bhatlawande, M. Manjunatha, J. Mukherjee, M. Biswas, D. Das, and S. Gupta (2014), Design, Development and Clinical Evaluation of the Electronic Mobility Cane for Vision Rehabilitation, IEEE Transactions on Neural Systems & Rehabilitation Engineering, 22(6), Nov., 1148-1159.
 

Remote Sensing  

A technique for classifying Landsat images using a novel feature of slope along the spectral band (named spectral slope) has been developed. The computation has been facilitated by GP/GPU processors under CUDA programming environment.

Related Publications

  • Shashaank M. Aswatha, Jayanta Mukhopadhyay, Prabir K. Biswas (2016): Spectral slopes for automated classification of land cover in landsat images. ICIP 2016: 4354-4358
  • Shashaank, M.A., Jayanta Mukherjee, P.K. Biswas,and S. Aikat (2017), Towards Automated Land Cover Classification in Landsat Images using Spectral Slopes at Different Bands, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 10(3), 1096-1104.
   

Computational Phylogenetics  

An efficient and fast algorithm for computation of supertree (called COPSEDTree) by analyzing the relations of pairs of taxa (couplets) has been developed. The algorithm significantly improves the storage complexity (quadratic in nature) compared to existing techniques. This makes it amenable to handle a large taxa-set. However, the tree it forms is not fully resolved. This has been further taken care of in an improved version of COSPEDTree has later been suggested. A few techniques to handle the problem of incomplete lineage sorting in predicting species tree from gene trees have also been developed.

Related Publications

  • Sourya Bhattacharyya, Jayanta Mukhopadhyay (2014): Couplet supertree by equivalence partitioning of taxa set and DAG formation. BCB 2014: 259-268
  • Sourya Bhattacharyya, Jayanta Mukherjee (2015): COSPEDTree: COuplet Supertree by Equivalence Partitioning of Taxa Set and DAG Formation. IEEE/ACM Trans. Comput. Biology Bioinform. 12(3): 590-603.
  • Sourya Bhattacharyya, Jayanta Mukhopadhyay (2015): Couplet Supertree Based Species Tree Estimation. ISBRA 2015: 48-59
  • Sourya Bhattacharyya, Jayanta Mukhopadhyay (2016): Accumulated Coalescence Rank and Excess Gene Count for Species Tree Inference. AlCoB 2016: 93-105
  • Sourya Bhattacharyya, Jayanta Mukhopadhyay (2016): COSPEDTree-II: Improved couplet based phylogenetic supertree. BIBM 2016: 98-101
 

Image inpainting  

A few techniques for image inpainting have been developed. It has been observed that different policies of prioritization lead to varying quality of reconstruction in different parts of the target region. An optimization criteria has been set to integrate results from inpainted images obtained by different initialization and prioritization schemes. A no-reference metric has also been used to check the quality of inpainted images. This has led to significant improvement in the quality of inpainted images. In another work image inpainting is applied for defencing of images.

Related Publications

  • Veepin Kumar, Jayanta Mukherjee, Shyamal Kumar Das Mandal (2015): Combinatorial Exemplar-Based Image Inpainting. IWCIA 2015: 284-298.
  • Veepin Kumar, Jayanta Mukhopadhyay, Shyamal Kumar Das Mandal (2015): Modified Exemplar-Based Image Inpainting via Primal-Dual Optimization. PReMI 2015: 116-125.
  • Veepin Kumar, Jayanta Mukherjee, Shyamal Kumar Das Mandal (2016): Image Inpainting Through Metric Labeling via Guided Patch Mixing. IEEE Trans. Image Processing 25(11): 5212-5226.
  • Veepin Kumar, Jayanta Mukherjee, Shyamal Kumar Das Mandal (2016): Image defencing via signal demixing. ICVGIP 2016: 11:1-11:8.
 

Compressed domain processing  

Different filtering algorithms in the block DCT domain have been developed. These are used in transcoding JPEG2000 images to JPEG and vice versa. A book on compressed domain image and video processing has been published, which include a few new algorithms for image filtering, resizing, color image enhancement, image saturation and desaturation in the DCT domain and color constancy.

Related Publications

  • J. Mukhopadhyay (2011): Image and Video Processing in the Compressed Domain, CRC Press.
  • Kapinaiah Viswanath, Jayanta Mukherjee, Prabir Kumar Biswas (2011): Image filtering in the block DCT domain using symmetric convolution. J. Visual Communication and Image Representation 22(2): 141-152.
  • Kapinaiah Viswanath, Jayanta Mukherjee, Prabir Kumar Biswas: Block DCT to wavelet transcoding in transform domain. Signal, Image and Video Processing 6(2): 179-195 (2012)
 

Medical Informatics and Telemedicine Systems  

A patient management system for Neonatal Intensive Care Unit (NICU) named NavajataK, has been developed. The system was deployed and tested in the Dept. of neonatology, Institute of Post Graduate Medical Education and Research (IPGMER). It has various novel components, including generation of an automated patient hand-over sheet, measurement of combination of drug and fluid feed required for neonates, follow-up of patients, etc.

Related Publications

  • S. Ray, Debi Prosad Dogra, S. Bhattacharya, B. Saha, Arunava Biswas, Arun K. Majumdar, Jayanta Mukherjee, Bandana Majumdar, Arun Kumar Singh, A. Paria, Suchandra Mukherjee, S. Das Bhattacharya (2011): A Web Enabled Health Information System for the Neonatal Intensive Care Unit (NICU). SERVICES 2011: 451-458
  • Debi Prosad Dogra, Karthik Nandam, Arun K. Majumdar, Shamik Sural, Jayanta Mukhopadhyay, Bandana Majumdar, Arun Kumar Singh, Suchandra Mukherjee (2012): A Tool for Automatic Hammersmith Infant Neurological Examination. IJEHMC 2(2): 1-13 (2011)
  • Soubhik Paul, Jayanta Mukhopadhyay, Arun K. Majumdar, Bandana Majumdar, S. Das Bhattacharya (2012): Methodology to visualize electronic health record for chronic diseases on small display screens. ICACCI 2012: 505-510
  • Arunava Biswas, Romil Roy, Sourya Bhattacharyya, Deepak Khaneja, Sangeeta Das Bhattacharya, Jayanta Mukhopadhyay (2016): Android application for therapeutic feed and fluid calculation in neonatal care - a way to fast, accurate and safe health-care delivery. BIBM 2016: 938-945