Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Subhasis ChaudhuriSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works245 from public data
- 647
- Spatially-Attentive Patch-Hierarchical Network for Adaptive Motion Deblurring336
This work proposes an efficient pixel adaptive and feature attentive design for handling large blur variations across different spatial locations and proposes an effective content-aware global-local filtering module that significantly improves performance by considering not only global dependencies but also by dynamically exploiting neighboring pixel information.
- Depth From Defocus: A Real Aperture Imaging Approach265
A Partial Derivatives of Various Quantities in CRB, a MAP-MRF approach to Depth Recovery and Restoration using MRF Models.
- Gait Analysis for Human Identification256
The dynamic time-warping (DTW) approach is used for matching so that non-linear time normalization may be used to deal with the naturally-occuring changes in walking speed.
- 226
- Off-line signature verification using DTW209
A signature verification system based on Dynamic Time Warping (DTW) that works by extracting the vertical projection feature from signature images and by comparing reference and probe feature templates using elastic matching is proposed.
- 191
- Spatially-Adaptive Image Restoration using Distortion-Guided Networks166
The key idea is to exploit the non-uniformity of heavy degradations in spatial-domain and suitably embed this knowledge within distortion-guided modules performing sparse normalization, feature extraction and attention.
- PIRM Challenge on Perceptual Image Enhancement on Smartphones: Report161
This paper reviews the first challenge on efficient perceptual image enhancement with the focus on deploying deep learning models on smartphones and proposes solutions that significantly improved baseline results defining the state-of-the-art for image enhancement on smartphones.
- Region-Adaptive Dense Network for Efficient Motion Deblurring148
This paper presents a new architecture composed of region adaptive dense deformable modules that implicitly discover the spatially varying shifts responsible for non-uniform blur in the input image and learn to modulate the filters, enabling almost real-time deblurring.
- 148
- 148
- 131
- 112
- An MRF model-based approach to simultaneous recovery of depth and restoration from defocused images106
The performance of the proposed MAP-Markov random field based scheme for recovering the depth and the focused image of a scene from two defocused images is found to be better than that of the existing window-based depth from defocus technique.
- AIM 2020 Challenge on Efficient Super-Resolution: Methods and Results101
The AIM 2020 challenge on efficient single image super-resolution was to super-resolve an input image with a magnification factor x4 based on a set of prior examples of low and corresponding high resolution images with focus on the proposed solutions and results.
- Bringing Alive Blurred Moments92
This work first learns motion representation from sharp videos in an unsupervised manner through training of a convolutional recurrent video autoencoder network that performs a surrogate task of video reconstruction that outperforms competing methods across all factors: accuracy, speed, and compactness.
- Human Face Detection in Cluttered Color Images Using Skin Color, Edge Information.80
A fast algorithm for detecting human faces in color images using color histogram for skin (in the HSV space) in conjunction with edge information to quickly locate faces in a given image is proposed.
- 75
- 74
- 73
- 69
- 68
- 68
- A variational approach to recovering depth from defocused images67
A smoothness constraint is imposed on the estimates of the blur parameter, and a variational approach to the DFD problem is developed using the space-frequency representation (SFR) framework.
- Analyzing Perception-Distortion Tradeoff Using Enhanced Perceptual Super-Resolution Network63
The proposed network, called enhanced perceptual super-resolution network (EPSR), is trained with a combination of mean squared error loss, perceptual loss, and adversarial loss and achieves the state-of-the-art trade-off between distortion and perceptual quality while the existing methods perform well in either of these measures alone.
- Unsupervised Class-Specific Deblurring63
An end-to-end deblurring network designed specifically for a class of data that learns a strong prior on the clean image domain using adversarial loss and maps the blurred image to its clean equivalent and imposes a scale-space gradient error with an additional gradient module.
- 62
- 59
- Resolution Enhancement of PMD Range Maps59
This work follows a Bayesian framework by modeling the original HR range as a Markov random field (MRF) and proposes the use of an edge-adaptive MRF prior to handle discontinuities.
- AIM 2019 Challenge on Image Demoireing: Methods and Results57
This paper reviews the first-ever image demoireing challenge that was part of the Advances in Image Manipulation (AIM) workshop, held in conjunction with ICCV 2019, and focuses on the proposed solutions and their results.
- 55
- 53
- Space-Variant Approaches to Recovery of Depth from Defocused Images53
The first approach proposed here models the DFD system as a block shift-variant one and incorporates the interaction of blur among neighboring subimages in an attempt to improve the estimate of the depth.
- AIM 2020 Challenge on Rendering Realistic Bokeh52
The proposed solutions significantly improved the baseline results, defining the state-of-the-art for practical bokeh effect rendering problem.
- 52
- Self-supervised Monocular Underwater Depth Recovery, Image Restoration, and a Real-sea Video Dataset50
- 50
- AIM 2020: Scene Relighting and Illumination Estimation Challenge50
The novel VIDIT dataset used in the AIM 2020 challenge and the different proposed solutions and final evaluation results over the 3 challenge tracks are presented.
- An Efficient Framework for Dense Video Captioning50
This paper proposes a deep reinforcement-based approach which enables an agent to describe multiple events in a video by watching a portion of the frames, and reduces the computational cost by processing fewer frames while maintaining accuracy.
- 49
- 45
- 45
- 44
- 43
- Range map superresolution-inpainting, and reconstruction from sparse data43
This work uses multiple relatively-shifted LR range images, where the motion between the LR images serves as a cue for super-resolution, and exploits a cue from segmentation of an optical image of the same scene, which constrains pixels in the same color segment to have similar range values.
- 43
- 41
- 38
- NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images: Methods and Results37
This paper reviews the NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images to establish a new and powerful benchmark for the task of removing raindrops under varying lighting and focus conditions.
- Robust and computationally efficient superresolution algorithm37
The proposed method not only preserves edges but also lends robustness to errors in the estimates of motion and blur parameters and uses theoretically the neighborhood structure for the posterior distribution in the presence of warping, blurring, and downsampling operations to effectively reduce the overall computations.
- 34
- 33
- Motion-free superresolution and the role of relative blur33
This work examines the theory of motion-free superresolution and derives analytical expressions in the discrete Fourier transform domain for obtaining a superresolved image from its blurred and downsampled versions and addresses the effect of relative blurring among the defocused observations on the quality of the reconstructed image.
- 32
- 32
- 31
- 31
- 30
- 30
- 29
- Inferring Image Transformation and Structure from Motion-Blurred Images29
A technique to estimate the scene depth with the knowledge of the estimated image transformations from observations that are blurred due to unconstrained camera motion is proposed.
- 28
- 28
- 27
- 27
- 27
- 27
- 27
- 25
- 23
- 22
- 21
- 21
- Mixed-dense connection networks for image and video super-resolution20
A deep architecture for single image super-resolution (SISR) is proposed, built using efficient convolutional units the authors refer to as mixed-dense connection blocks (MDCB), which combines the strengths of both residual and dense connection strategies, while overcoming their limitations.
- Depth inpainting by tensor voting19
This paper proposes a fast and reliable algorithm for depth map inpainting using the tensor voting (TV) framework, and align the depth maps of the training set with the target (defective) depth map and evaluates the goodness of depth estimates among candidate values using 3D TV.
- 18
- Multilevel weighted enhancement for underwater image dehazing18
An approach for single-image dehazing by multilevel weighted enhancement of the image that reveals that their linear combination along with the coarsest information can successfully restore the image.
- Shape from Sharp and Motion-Blurred Image Pair17
This work develops a technique to estimate the transformation spread function (TSF) which symbolizes the camera shake and recovers the complete depth map of the scene within a regularization framework using an unblurred–blurred image pair.
- Towards Unrestrained Depth Inference with Coherent Occlusion Filling17
This work proposes a depth estimation framework using calibrated images captured under general camera motion and lens parameter variations that seeks to generalize the constrained areas of stereo and shape from defocus (SFD)/focus (SFF) by handling, in tandem, various effects such as focus variation, zoom, parallax and stereo occlusions, all under one roof.
- 17
- Face recognition using multiple facial features17
A face recognition method that fuses information acquired from global and local features of the face for improving performance is proposed and validated on FERET and AR databases.
- A recursive algorithm for maximum likelihood-based identification of blur from multiple observations17
A maximum likelihood-based method is proposed for blur identification from multiple observations of a scene when the relations among the blurring functions are known and the estimate of blur obtained is very good.
- 16
- MRF model-based identification of shift-variant point spread function for a class of imaging systems16
The proposed MAP-MRF based scheme for identifying a class of shift-variant imaging systems whose point spread function can be parameterized by a single blur parameter is proposed and the estimates of the SV blur are found to be better than that of the window-based blur identification technique.
- Performance Analysis of Maximum Likelihood Estimator for Recovery of Depth from Defocused Images and Optimal Selection of Camera Parameters16
A criterion for optimal selection of camera parameters to obtain an improved estimate of the depth is proposed and is based on the Cramer-Rao bound of the variance of the error in the estimate of blur.
- 15
- 15
- 15
- 14
- 14
- Shape extraction of low‐textured objects in video microscopy13
This paper exploits spatial dependencies by modeling the shape of the object with a discontinuity‐adaptive Markov random field wherein the focus measure profile is used to judiciously control the degree of smoothness.
- Restoration of scanned photographic images13
Based on the one-dimensional particle filter, a new approach is proposed for restoration under space-invariant as well as space-variant blurring conditions by propagating the samples of the probability distribution through an appropriate state model.
- 13
- 12
- 12
- Joint image and depth completion in shape-from-focus: Taking a cue from parallax12
The interesting possibility of exploiting motion parallax cue in the images captured in Shape-from-focus with a practical camera to jointly inpaint the focused image and depth map is demonstrated.
- Scale-Recurrent Multi-residual Dense Network for Image Super-Resolution11
A scale recurrent SR architecture built upon units containing series of dense connections within a residual block (Residual Dense Blocks) that allow extraction of abundant local features from the image while being parametrically more efficient as compared to current state-of-the-art approaches.
- Generating high quality pan-shots from motion blurred videos11
A method to generate a pan photo from a video captured using a hand-held camera that first segments out the moving object by motion compensation of the background and then estimates the inter-frame velocity and relative depth of the object.
- A Model-Based Approach to Shape from Focus.11
A lithographic silver halide photographic light-sensitive material is described which scarcely exhibits colored fringe and black pepper, forms halftone dots of good quality whether a fresh or an exhausted developer is used for processing thereof.
- Recursive framework for joint inpainting and de-noising of photographic films10
A recursive image recovery scheme based on the unscented Kalman filter (UKF) to simultaneously inpaint identified damaged portions in an image and suppress film-grain noise is presented.
- 10
- 9
- 9
- 9
- Discontinuity-Adaptive Shape from Focus Using a Non-convex Prior9
A new SFF method that can provide rich structure information given limited number of observations and pose the shape estimation as a minimization problem within a Maximum A Posteriori - Markov Random Field framework.
- 9
- Quantitative analysis of error bounds in the recovery of depth from defocused images9
A new theorem is proposed that proves that the Cramér-Rao bound on the variance of the error in the estimate of blur decreases with an increase in the number of observations, and one can indeed get better estimates of depth from multiple defocused images compared with those using only a single image.
- 8
- 8
- Spatially-Adaptive Residual Networks for Efficient Image and Video Deblurring8
Extensive qualitative and quantitative comparisons with prior art on benchmark dynamic scene deblurring datasets clearly demonstrate the superiority of the proposed networks via a reduction in model-size and significant improvements in accuracy and speed, enabling almost real-timedeblurring.
- 8
- 8
- 7
- Deep end-to-end rolling shutter rectification7
The end-to-end formulation in this model does not constrain the estimated motion to ground truth motion parameters, thereby successfully rectifying the RS images with complex real-life camera motion.
- Robust Super-Resolution of Real Faces Using Smooth Features7
A two module super-resolution network where the feature extractor module extracts robust features from the LR image, and the SR module generates an HR estimate using only these robust features.
- DEPTH INPAINTING WITH TENSOR VOTING USING LOCAL GEOMETRY7
This paper proposes a fast and simple algorithm for range map inpainting using Tensor Voting (TV) framework and uses 3D tensor voting for estimating different plane equations and pass depth estimates from all possible local planes that pass through a missing region.
- HDR Imaging under Non-uniform Blurring7
A technique to obtain the high dynamic range (HDR) irradiance of a scene from a set of differently exposed images captured using a hand-held camera and a transformation spread function (TSF) that represents space-variant blurring as a weighted average of differently transformed versions of the latent image.
- 7
- Inpainting in Multi-image Stereo7
This work proposes a new method for inpainting both image and depth of a scene using multiple stereo images and establishes the approaches in the belief propagation (BP) framework which also uses the segmentation cue for estimation/inpainting of depth maps.
- Depth estimation with a practical camera7
This work proposes a framework for depth estimation from a set of calibrated images, captured under general camera motion and parameter variation, and uses the vital cue from color image segmentation to constrain the estimation process.
- 7
- Extension of the shape from focus method for reconstruction of high-resolution images7
This work proposes an extension to the traditional SFF method to optimally estimate a high-resolution image of the 3D object, given the low-resolution observations and the depth map derived from traditional S FF, and uses the observation stack to achieve significant improvement in resolution.
- A Discontinuity Adaptive Method for Super-Resolution of License Plates7
A super-resolution algorithm tailored to enhance license plate numbers of moving vehicles in real traffic videos is proposed and a discontinuity adaptive regularizer is proposed to preserve edges in the reconstructed number plate for better readability.
- 7
- Image recovery under nonlinear and non-Gaussian degradations7
A new two-dimensional recursive filter for recovering degraded images is proposed that is based on particle-filter theory and validated by demonstrating its effectiveness in recovering images degraded by film-grain noise.
- 7
- Higher-order-statistics-based detection of vehicles in still images7
When tested on real aerial images, the method gives good results, even for complicated scenes, while the false alarms are very few, and can serve as an important step toward building an automated traffic monitoring system.
- Locating Human Faces in a Cluttered Scene7
Two new schemes for finding human faces in a photograph using a distribution-based model approach and a hidden Markov model (HMM) based approach to the face-finding problem are presented.
- 6
- 6
- Real Aperture Axial Stereo: Solving for Correspondences in Blur6
This work proposes an algorithm which fuses the twin cues of defocus and parallax for recovering 3D structure and shows that in real-aperture axial stereo, both blur and pixel motion are tightly coupled to the underlying shape of the object.
- Harnessing defocus blur to recover high-resolution information in shape-from-focus technique6
The authors propose to exploit the defocus information in the stack of LR images to obtain a super-resolved image as well as a high-resolution (HR) depth map of the underlying 3D object.
- 6
- Robust space-variant super-resolution6
This paper uses a maximum a posteriori-Markov random field (MAP-MRF) based approach for recovering the high-resolution image and shows that modeling by MRF lends robustness to errors in estimation of motion and blur parameters.
- Image restoration using the particle filter: handling non-causality6
A recursive-state estimation scheme for image restoration using the particle filter with a novel choice for the state vector and a concurrent block estimation technique to incorporate full-plane regions of support for the image model as well as the blur.
- Depth Recovery from Defocused Images6
In machine vision, most of the research on recovering scene structure is based on a pinhole camera model, but practical camera systems (and also the human eye) are not pinhole cameras but consist of real aperture lenses.
- 5
- Color restoration in turbid medium5
This work proposes a color modulation scheme that is applied prior to color mapping to accommodate the same mapping function for different depths as well and results are given on several images to validate the efficacy of the proposed methodology.
- 5
- Inpainting in Shape from Focus: Taking a Cue from Motion Parallax5
This paper investigates the effect of motion parallax in SFF and demonstrates the interesting possibility of how it can be judiciously used to jointly inpaint image and depth profiles.
- 5
- 4
- Unpaired Image Denoising4
First, a flow-based generative model is used to learn a prior from clean images, which is then used to train a denoising network without the need for any clean targets.
- 4
- 4
- Deskewing by space-variant deblurring4
The notion of virtual depth map is introduced which is assigned to a planar scene when observed through a dynamic water surface and transform the deskewing problem into one of space-variant deblurring from a single image within an alternating minimization framework.
- Deep Decoupling of Defocus and Motion Blur for Dynamic Segmentation4
A deep convolutional neural network is developed to predict the probabilistic distribution of the composite kernel which is the convolution of motion blur and defocus kernels at each pixel, which enables jointly handling defocus and motion blur to resolve depth-motion ambiguity.
- 4
- 4
- Image matching with higher-order geometric features4
A geometric matching technique in which line segments and elliptical arcs are used as edge features and the use of these higher-order features renders feature representation efficient.
- 4
- 4
- Gait-Based Human Identification Using Appearance Matching4
In this chapter, an appearance-based approach for recognizing human gait is presented and dynamic time-warping (DTW), which is a nonlinear time-normalization technique, is used to deal with naturally occurring changes in the walking speeds of individuals.
- 4
- Image clustering using higher-order statistics4
A more general higher-order statistics-based closeness measure derived from a series expansion for a multivariate probability density function in terms of the Gaussian function and the Hermite polynomials is proposed for clustering.
- 3
- PNeRV: A Polynomial Neural Representation for Videos3
Polynomial Neural Representation for Videos (PNeRV), a parameter-wise efficient, patch-wise INR for videos that preserves spatiotemporal continuity, and a custom Hierarchical Patch-wise Spatial Sampling Scheme that ensures spatial continuity while retaining parameter efficiency.
- 3
- 3
- Mosaicing deep underwater imagery3
This work proposes a framework that restores deep UW images captured by Remotely Operated Vehicles in accordance with a suitably derived degradation model and harnesses the scene geometry information present in each image to aid in constructing a mosaic that is free from artifacts such as local blurring, ghosting, double contouring and visible seams.
- 3
- Blind restoration of aerial imagery degraded by spatially varying motion blur3
This paper develops a methodology for blind restoration of spatially varying blur induced by camera motion caused by instabilities of the moving platform and proposes a scheme to estimate the original focused image affected by arbitrarily-shaped blur kernels.
- 3
- 3
- Unified multiframe super-resolution of matte, foreground, and background3
This paper proposes a multiframe approach to increase the spatial resolution of the matte, foreground, and background and shows that joint estimation is advantageous, as super-resolved edge information helps in obtaining a sharp matte, while the matte in turn aids in resolving fine details.
- 3
- 3
- A New Extension of Kalman Filter to Non-Gaussian Priors3
This paper proposes a sampling-based Kalman Filter (KF) for the image estimation problem that compute the mean and covariance of the prior (possibly non-Gaussian) by importance sampling and shows that the estimates obtained are superior to the traditional Kalman filter that uses the auto-regressive state model.
- Motion-Free Superresolution.3
The theory of motion-free superresolution is examined by formulating the problem in the Discrete Time Fourier Transform (DTFT) and the DFT domain and provides some new insights into how aliasing and blurring can effect exact reconstruction of the superresolved image.
- Data clustering using hierarchical deterministic annealing and higher order statistics [image processing]3
Experimental results clearly demonstrate the improvement in the performance of the HDA algorithm when the interdependency among the clusters and the HOS of the data points are also utilized for the purpose of clustering.
- Single Noisy Image Super Resolution by Minimizing Nuclear Norm in Virtual Sparse Domain2
The approach minimizes nuclear norm in a virtual sparse domain to explore the inherent low-rank structure of visual data, and is further augmented with coarse-to-fine information by adaptively re-aligning the data along the principal components of a dictionary invirtual sparse domain.
- Recognizing blurred, nonfrontal, illumination, and expression variant partially occluded faces2
The focus of this paper is on the problem of recognizing faces across space-varying motion blur, changes in pose, illumination, and expression, as well as partial occlusion, when only a single image per subject is available in the gallery.
- 2
- 2
- 2
- Illumination robust change detection with CMOS imaging sensors2
An optimization problem where each row of the distorted image is approximated as a weighted sum of the corresponding rows in warped versions of the reference image due to camera motion within the exposure period to account for geometric as well as photometric differences is formulated.
- Depth Estimation and Inpainting with an Unconstrained Camera2
This work proposes a depth estimation approach by elegantly coupling the motion and defocus cues, and uses the belief propagation method in the estimation approach, which also handles occlusions and using the color image segmentation cue.
- 2
- 2
- An efficient representation of digital curves with line segments and elliptical arcs2
A new method to measure the curvature at a point on the curve, which is based on calculating the rate of change of tangential angle with respect to the curve length is proposed.
- 2
- Robust Face Recognition by Fusion.2
A novel probabilistic framework that combines information acquired from different facial features for robust face recognition is proposed and the match with the highest score is used to establish the identity of the person.
- 2
- 2
- Passive Methods for Depth Recovery2
The apparatus available to support depth estimation in machine vision can extend beyond the cues of human vision, and in difficult circumstances, a combination of these cues may be required to obtain reliable depth information.
- 1
- 1
- Semi-Supervised Implicit Augmentation for Data-Scarce VQA1
SemIAug is introduced, a model and dataset agnostic strategy specially designed to address the challenges faced by limited data availability in the domain-specific VQA datasets and improves the VQA performance on data-scarce datasets without the need for additional data or labels.
- Unsupervised network for low-light enhancement1
An unsupervised low-light enhancement network using contextguided illumination-adaptive norm (CIN) and a region- Adaptive single input multiple output (SIMO) model that can generate multiple enhanced images from a single lowlight image.
- Latents2Semantics: Leveraging the Latent Space of Generative Models for Localized Style Manipulation of Face Images1
The Latents2Semantics Autoencoder (L2SAE), a Generative Autoencoder model that facilitates highly localized editing of style attributes of several Regions of Interest (ROIs) in face images, is introduced, allowing for structure-preserving style editing of the chosen ROIs.
- 1
- Latents2Segments: Disentangling the Latent Space of Generative Models for Semantic Segmentation of Face Images1
The endeavour in this work is to do away with the priors and complex pre-processing operations required by SOTA multi-class face segmentation models by reframing this operation as a downstream task post infusion of disentanglement with respect to facial semantic regions of interest (ROIs) in the latent space of a Generative Autoencoder model.
- 1
- Multi-planar geometry and latent image recovery from a single motion-blurred image1
This work proposes a novel approach to estimate the normal of a planar surface from a single motion-blurred image of a multi-planar 3D scene and develops an algorithm for automatic recovery of the number of planes, the parameters corresponding to each plane, and camera motion from asingle motion- Blurred image.
- 1
- Planar Geometry and Latest Scene Recovery from a Single Motion Blurred Image1
An approach for estimation of normal of a planar scene from a single motion blurred observation and a first-of-its-kind approach to recover the planar geometry and latent image of the scene by adopting an alternating minimization framework built on the findings.
- A Methodology to Reconstruct Large Damaged Regions in Heritage Structures1
This chapter discusses geometric reconstruction of large damaged regions (or holes) in 3D digital models and uses two methodologies to effectively harness the geometric prior: a non-iterative framework based on tensor voting when multiple self-similar examples are available and a dictionary learning-based method when only a singleSelf-similar example is available.
- 1
- Framework for Processing Videos in the Presence of Spatially Varying Motion Blur1
It must be emphasized that motion blur can also serve as a vital cue for camera motion estimation, depth recovery, super-resolution, and image forensics.
- Change detection in underwater imagery1
This work proposes a distort-difference pipeline that employs an alternating minimization framework to perform change detection in the presence of geometric distortions (skew) as well as photometric degradations (blur and global illumination variations).
- 1
- Camouflaging Motion Blur1
This work applies a suitably derived blurring model to smear a target object and naturally blend it into the motion blurred background by using alpha matting to raise the interesting issue of camouaging motion blur with special relevance to non-uniformly blurred images.
- 1
- Matte Super-Resolution for Compositing1
This paper treats super-resolution of the matte and foreground in a combined framework, incorporating the matting equation in the image degradation model, and uses a discontinuityadaptive smoothness prior to include observed data in the solution.
- 1
- Robust Face Recognition in the Presence of Clutter1
A new method within the framework of principal component analysis to robustly recognize faces in the presence of clutter is proposed and a suitable classifier is derived to distinguish non-face patterns from faces.
- MRF Model-Based Identification of Shift-Variant PSF1
The problem addressed by Pentland Pen87 is a special case of the more general DFD problem in which the depth must be recovered given two defocused images of a scene, and effectively boils down to identifying a space-variant Gaussian defocusing process, given the input focused image and the degraded output image.
- –
- –
- Spatially-Attentive Patch-Hierarchical Network with Adaptive Sampling for Motion Deblurring–
A pixel adaptive and feature attentive design for handling large blur variations across different spatial locations and process each test image adaptively and introduces a pixel-adaptive non-uniform sampling strategy that implicitly discovers the difficult-to-restore regions present in the image and performs fine-grained refinement in a progressive manner.
- Zero shot framework for satellite image restoration–
A mechanism to disentangle distortion is proposed and the use of knowledge distillation to train a restoration network using the generated image pairs and ablation studies show that the proposed mechanism successfully disentangles distortion.
- Unsupervised haze removal from underwater images–
This work proposes a new method to remove haze from underwater images using unpaired data using a Haze Disentanglement Network (HDN) and shows that the results of this method outperform prior art both visually and quantitatively.
- –
- Exploring the Effectiveness of Mask-Guided Feature Modulation as a Mechanism for Localized Style Editing of Real Images–
The SemanticStyle Autoencoder (SSAE), a deep Generative Autoen coder model that leverages semantic mask-guided latent space manipulation for highly localized photorealistic style editing of real images is presented.
- –
- –
- –
- Real-time multiple point target detection and tracking in infrared imagery✱–
This work addresses the problem of real-time detection and tracking of multiple point targets in diverse background conditions by applying a ’top-hat’ operator as the first stage of the detection algorithm and followed by a systematic thresholding scheme to yield a good balance between MD and FA.
- –
- –
- –
- –
- –
- –
- –
- –
- Color Image Super Resolution in Real Noise–
This paper proposes to super-resolve noisy color images by considering the color channels jointly through regularization term that involves projection onto PCA basis, which is constructed using similar patches extracted across different scales of the input image.
- –
- Dictionary Replacement for Single Image Restoration of 3D Scenes.–
This paper addresses the problem of jointly estimating the latent image and the depth/blur map from a single space-variantly blurred image using dictionary learning using a multi-label MRF and solves it using graph-cut.
- –
- Cueing motion blur for registration of inclined planar scenes–
This work extends the state-of-the-art by cueing motion blur itself to infer plane inclination by matching extremities of blur kernels computed at different locations in the image by showing that it is possible to find the orientation.
- Layered Depth Inference from Flow in Underwater Imagery–
Under reasonable assumptions, it is established that the motion blur induced by commonly observed uid flows can serve as a valuable cue for inferring the underlying depth layers of the scene.
- –
- –
- –
- Editorial introduction to the special issue–
Extended versions of six papers that were first presented at the Seventh Indian Conference on Computer Vision, Graphics, and Image Processing (ICVGIP) are brought to you, in this special issue.
- –
- Space-Variant Filtering Models for Recovering Depth–
A new approach to time-variant filtering is suggested, based upon the generation of the time-frequency representation (TFR) of a signal, multiplication of that representation by a function H(w,t), which can be regarded as theTime-varying transfer function, and obtaining a filtered output by an inverse operation, assuming that the inverse exists.
- –
- ML Estimation of Depth and Optimal Camera Settings–
In Section 2.2.3, it was pointed out that the recovery of depth from defocused images, in its generality, is equivalent to the problem of space-variant blur identification.
- Recursive Computation of Depth from Multiple Images–
In this chapter, it was theoretically proved that the use of additional observations, if available, can lead to improved estimates of the depth of a scene.
- –
- Depth Recovery with a Block Shift-Variant Blur Model–
This work has shown that splitting an image into independent subimages introduces an error in the estimate of blur due to improper selection of boundary, which can be reduced by choosing a larger size for the subimage.
- Simultaneous Depth Recovery and Image Restoration–
This work proposes a multiple model-based extended Kalman filter for restoration of spatially varying blurred images and proposes the use of projections onto convex sets for space-varying image restoration.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-15. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.