Showing papers by "Sudhish N. George published in 2016"

Reconstruction of cloud-contaminated satellite remote sensing images using kernel PCA-based image modelling

Abstract: The presence of clouds can restrict the potential uses of remote sensing satellite imagery in extracting information and interpretation. Automatic detection and removal of clouds which hide significant information in the image is an important task in remote sensing. Hence, our aim is to detect clouds and restore the missing information in order to make the image ready for further analysis and applications. Due to the difference in nature and appearance, thick and thin clouds are dealt separately. Thick cloud is detected using an efficient Fuzzy C-Means (FCM) clustering algorithm, while thin cloud is detected using a simple region growing technique. In order to reconstruct the missing pixels, we utilize the prior knowledge about the statistics of the specific image class. Kernel principal component analysis (KPCA)-based image model is obtained using a set of training images. Missing area in the image is restored after an iterative projection operation and gradient descent algorithm. In short, an image lying out of the modelled image space is iteratively modified to obtain the restored image and that would be in the image space according to the obtained nonlinear low-dimensional and sparse KPCA image model. To illustrate the performance of the proposed method, a thorough experimental analysis on FORMOSAT multi-spectral images is done using MATLAB platform. When compared to the two recent existing techniques, our proposed method is superior and makes a promising tool for thick and thin cloud removal in multi-spectral satellite images.

A novel approach for de-speckling of ultrasound images using bilateral filter

Abstract: Ultrasound images are degraded due to a multiplicative noise also known as speckle which not only deteriorates the edges but also brings about unfaithful indications about the nature of the image. Ideal de-speckling algorithms should preserve the edges as well as the texture of the image. In the proposed scheme, the frequency components in the image are obtained using wavelet decomposition and the noise in the high frequency components is estimated by incorporating a bayesian framework. The low frequency noise components are de-speckled by passing through a guided bilateral filter. The proposed scheme was compared with some existing works which include Bayesian non local means filter (OBNLM), wavelet estimation based on non-parametric model and fast bilateral filtering. Quantitative assessments proved that the proposed scheme performed better both in edge preservation and visual quality. Experiments were conducted both on natural images and ultrasound images and both gave good results.

Modified dictionary learning method for sparsity based single image super-resolution

Abstract: This paper proposes a self-learning based dictionary training approach for single image super-resolution where the low-resolution (LR) and high-resolution (HR) dictionaries are jointly trained using training data set and different scaled versions of the input image. Local variance based salient feature identification is carried out to speed up the super-resolution algorithm. It can be demonstrated that our modified SR algorithm can qualitatively and quantitatively outperform bicubic interpolation and state-of-the-art methods.

Source Separation of Heartbeat Sounds for Effective E-Auscultation

Abstract: This paper proposes a cost effective solution for improving the effectiveness of e-auscultation. Auscultation is the most difficult skill for a doctor, since it can be acquired only through experience. The heart sound mixtures are captured by placing the four numbers of sensors at appropriate auscultation area in the body. These sound mixtures are separated to its relevant components by a statistical method independent component analysis. The separated heartbeat sounds can be further processed or can be stored for future reference. This idea can be used for making a low cost, easy to use portable instrument which will be beneficial to people living in remote areas and are unable to take the advantage of advanced diagnosis methods.