Mithun Das Gupta

Non-negative matrix factorization as a feature selection tool for maximum margin classifiers

Non-negative matrix factorization (NMF) has previously been shown to be a useful decomposition tool for multivariate data. Non-negative bases allow strictly additive combinations which have been shown to be part-based as well as relatively sparse. We pursue a discriminative decomposition by coupling NMF objective with a maximum margin classifier, specifically a support vector machine (SVM). Conversely, we propose an NMF based regularizer for SVM. We formulate the joint update equations and propose a new method which identifies the decomposition as well as the classification parameters. We present classification results on synthetic as well as real datasets.

Bi-affinity filter: a bilateral type filter for color images

We propose a new filter called Bi-affinity filter for color images. This filter is similar in structure to the bilateral filter. The proposed filter is based on the color line model, which does not require the explicit conversion of the RGB values to perception based spaces such as CIELAB. The bi-affinity filter measures the affinity of a pixel to a small neighborhood around it and weighs the filter term accordingly. We show that this method can perform at par with standard bilateral filters for color images. The small edges of the image are usually enhanced leading to a very easy image enhancement filter.

Restoration and recognition in a loop

In this paper we present a novel learning based method for restoring and recognizing images of digits that have been blurred using an unknown kernel. The novelty of our work is an iterative loop that alternates between recognition and restoration stages. In the restoration stage we model the image as an undirected graphical model over the image patches with the compatibility functions represented as non-parametric kernel densities. Compatibility functions are initially learned using uniform random samples from the training data. We solve the inference problem by an extended version of the non-parametric belief propagation algorithm in which we introduce the notion of partial messages. We close the loop by using the confidence scores of the recognition to non-uniformly sample from the training set in order to retrain the compatibility functions. We show experimental results on synthetic and license plate images.

Non-parametric image super-resolution using multiple images

In this paper, we present a novel learning based framework for performing super-resolution using multiple images. We model the image as an undirected graphical model over image patches in which the compatibility functions are represented as non-parametric kernel densities which are learnt from training data. The observed images are translation rectified and stitched together onto a high resolution grid and the inference problem reduces to estimating unknown pixels in the grid. We solve the inference problem by using an extended version of the non-parametric belief propagation algorithm. We show experimental results on synthetic digit images and real face images from the ORL face dataset.

Models for patch-based image restoration

We present a supervised learning approach for object-category specific restoration, recognition, and segmentation of images which are blurred using an unknown kernel. The novelty of this work is a multilayer graphical model which unifies the low-level vision task of restoration and the high-level vision task of recognition in a cooperative framework. The graphical model is an interconnected two-layer Markov random field. The restoration layer accounts for the compatibility between sharp and blurred images and models the association between adjacent patches in the sharp image. The recognition layer encodes the entity class and its location in the underlying scene. The potentials are represented using nonparametric kernel densities and are learnt from training data. Inference is performed using nonparametric belief propagation. Experiments demonstrate the effectiveness of our model for the restoration and recognition of blurred license plates as well as face images.

Models for Patch Based Image Restoration

In this paper we present a supervised learning approach for object-category specific restoration, recognition and segmentation of images which are blurred using an unknown kernel. The feature of this work is a multi layer graphical model which unifies the low level vision task of restoration, and the high level vision task of recognition in a cooperative framework. Proposed graphical model is an interconnected two layer Markov Random Field. The restoration layer accounts for the compatibility between sharp and blurred patches, and models the association between adjacent patches in the sharp image. The recognition layer encodes the patch location and class. The potentials are represented using non-parametric kernel densities and are learnt from the training data. Inference is performed using nonparametric belief propagation. We propose a similar model for super-resolution from multiple frames, and suggest the use of ordinal regression for sub-pixel shift estimation to address the registration issues. Experiments demonstrate the effectiveness of proposed models for the restoration and recognition of blurred license plate and face images.

Non-convex P-Norm Projection for Robust Sparsity

In this paper, we investigate the properties of Lp norm (p ≤1) within a projection framework. We start with the KKT equations of the non-linear optimization problem and then use its key properties to arrive at an algorithm for Lp norm projection on the non-negative simplex. We compare with L1 projection which needs prior knowledge of the true norm, as well as hard thresholding based sparsification proposed in recent compressed sensing literature. We show performance improvements compared to these techniques across different vision applications.

Simultaneous Registration and Segmentation by L1 Minimization

This paper studies the problem of simultaneously registering and segmenting a pair of images despite the presence of non-smooth boundaries. We assume one of the images is well segmented by automated algorithm or user interaction. This image acts as an atlas for the segmentation process. For the remaining images in the set, we propose a novel L1 minimization based technique, which leverages the fact that the other image is not closely segmented but has a reasonably ’thin’ boundary around it. The images are allowed to have non-rigid transformations amongst each other. We extend the two image formulation to multiple image registration and segmentation by introducing a low rank prior on the error matrix. We compare against rigid as well as non-rigid registration techniques. We present results on multi-modal real medical data.

Random forest based erythema grading for psoriasis

Psoriasis Area and Severity Index, or PASI score [7] is one of the most prevalent scoring indices for Psoriasis. Erythema or redness of skin is an important identifier for evaluation of PASI score. Extra subjectiveness in the evaluation of erythema has been observed, since the perception of redness can be influenced by the skin tone, ambient lighting and many other such factors which are difficult to control in a clinical setting. We propose a novel colorimetric feature for erythema grading by extending the tissue-photon interaction model [12] to make it skin tone independent. We propose to use Skellam distribution statistics as feature vectors for erythema grading. We present a random forest based technique for classification of erythema regions in Psoriasis into severe, moderate and slight categories.

Quality metric for parasternal long axis b-mode echocardiograms

This paper presents a method for automatically estimating the quality of Parasternal Long AXis (PLAX) B-mode echocardiograms. The purpose of the algorithm is to provide live feedback to the user on the quality of the acquired image. The proposed approach uses Generalized Hough Transform to compare the structures derived from the incoming image to a representative atlas, thereby providing a quality metric (PQM). On 133 PLAX images from 35 patients, we show: 1) PQM has high correlation with manual ratings from an expert echocardiographer 2) PQM has high correlation with contrast-to-noise ratio, a traditional indicator of image quality 3) on images with high PQM, error in automatic septal wall thickness measurement is low, and vice versa.