Abstract

This paper deals with the tracking and following of a person with a camera mounted mobile robot. A modified energy based optical flow approach is used for motion segmentation from a pair of images. Further a spatial relative veolcity based filering is used to extract prominently moving objects. Depth and color information are also used to robustly identify and follow a person.

Abstract

Current denoising techniques use the classical orthonormal wavelets for decomposition of an image corrupted with additive white Gaussian noise, upon which various thresholding strategies are built. The use of available biorthogonal wavelets in image denoising is less common because of their poor performance. In this paper, we present a method to design image-matched biorthogonal wavelet bases and report on their potential for denoising. We have conducted experiments on various image datasets namely Natural, Satellite and Medical with the designed wavelets using two existing thresholding strategies. Test results from comparing the performance of matched and fixed biorthogonal wavelets show an average improvement of 35% in MSE for low SNR values (0 to 18db) in every dataset. This improvement was also seen in the PSNR and visual comparisons. This points to the importance of matching when using wavelet-based denoising

Abstract

Retinal images are widely used to manually or automatically detect and diagnose many diseases. Due to the complex imaging setup, there is a large luminosity and contrast variability within and across images. Here, we use the knowledge of the imaging geometry and propose an enhancement method for colour retinal images, with a focus on contrast improvement with no introduction of artifacts. The method uses non-uniform sampling to estimate the degradation and derive a correction factor from a single plane. We also propose a scheme for applying the derived correction factor to enhance all the colour planes of a given image. The proposed enhancement method has been tested on a publicly available dataset [8]. Results show marked improvement over existing methods.

Abstract

Extraction of anatomical structures (landmarks), such as optic disk (OD), fovea and blood vessels, from fundus images is useful in automatic diagnosis. Current approaches largely use spatial relationship among the landmarks’ position for detection. In this paper, we present an appearance-based method for detecting fovea and OD from colour images. The strategy used for detection is based on improving the local contrast which is achieved by combining information from two spectral channels of the given image. The proposed method has been successfully tested on different datasets and the results show 96% detection for fovea and 91% detection for OD (a total of 502 and 531 images for fovea and OD are taken respectively).

Abstract

We present a new method for optic disk (OD) detection in a retinal image. It is an hybrid approach which uses properties of both appearance and model-based approaches which are mostly used for OD detection. An extrema pyramidal decomposition is employed and hill type topographical features are extracted at the lowest level of the image pyramid. The detected hill points are characterised as candidate OD locations. Later, a confidence measure is derived for each candidate using vessel structure information and candidate with the highest value is declared as final OD location. This method has been tested on different retinal image dataset and quantitative results are presented

Abstract

In this paper we present a method for fovea localization which does not use organization information of other retinal structures like optic disk and arcades. The main advantage of this method is that it does not require segmentation/localization of other retinal structures which are required as prior knowledge in existing fovea localization methods. The key idea is to enhance the relative contrast between the fovea and its surrounding such that it is well-separated in a retinal image. The approach was tested on 520 retinal images which include normal and pathological color retinal images. An overall accuracy of 90.57% is reported independent of left eye, right eye, normal or pathological retinal images.

Abstract

Capillary Non-Perfusion (CNP) is a condition in diabetic retinopathy where blood ceases to flow to certain parts of the retina, potentially leading to blindness. This paper presents a solution for automatically detecting and segmenting CNP regions from fundus fluorescein angiograms (FFAs). CNPs are modelled as valleys, and a novel multiresolution technique for trough-based valley detection is presented. The proposed algorithm has been tested on 40 images and validated against expert-marked ground truth. Obtained results are presented as a receiver operating characteristic (ROC) curve. The area under this curve is 0.842 and the distance of ROC from the ideal point (0,1) is 0.31.

Abstract

Automatic identification of a script in a given document image facilitates many important applications such as automatic archiving of multilingual documents, searching online archives of document images and for the selection of script specific OCR in a multilingual environment. In this paper, we model script identification as a texture classification problem and examine a global approach inspired by human visual perception. A generalised, hierarchical framework is proposed for script identification. A set of energy and intensity space features for this task is also presented. The framework serves to establish the utility of a global approach to the classification of scripts. The framework has been tested on two datasets: 10 Indian and 13 world scripts. The obtained accuracy of identification across the two datasets is above 94%. The results demonstrate that the framework can be used to develop solutions for script identification from document images across a large set of script classes.

Abstract

Unsupervised methods for automatic vessel segmentation from retinal images are attractive when only small datasets, with associated ground truth markings, are available. We present an unsupervised, curvature-based method for segmenting the complete vessel tree from colour retinal images. The vessels are modeled as trenches and the medial lines of the trenches are extracted using the curvature information derived from a novel curvature estimate. The complete vessel structure is then extracted using a modified region growing method. Testresults of the algorithm using the DRIVE dataset are superior to previously reported unsupervised methods and comparable to those obtained with the supervised methods.

Abstract

This paper presents a multi-component; distributed system (MuDiS) based solution for building a virtual learning environment which combines a wide range of technology, tools and digital gaming concepts to create an interactive tool for science education. MuDiS has been designed to be an extensible and easy-to-use system. The proposed environment is intended for designing and monitoring of educational content as well as creating a platform for exploring ideas. The system allows exchange of educational content and integrate different pedagogical approaches to learning and teaching under the same environment. We have developed a virtual physics lab that serves as an exemplar for the proposed virtual environment. We present the design details of the physics lab and discuss its performance.