Arun A. Balakrishnan

Implementation of radix-2 and split-radix fast fourier transform algorithm using current mirrors

Implementation of radix-2 and split-radix fast Fourier transform (FFT) algorithm using analog CMOS current mirrors with a reduction in the count of transistors and propagation delay is presented. The proposed method reduces the number of transistors required to implement analog FFTs and the percentage reduction increases considerably for higher point FFTs. It is shown that the number of transistors needed can be reduced further by using the split-radix FFT algorithm for analog implementation. This method decreases the computational delay since transistor count in the critical path gets reduced significantly for higher order FFTs. The work includes simulation and verification of generalized 4-point and 8-point FFT using TANNER EDA tools in 1.25 µm CMOS process. For 256-point FFT, the proposed method has a reduction of 14.76% for radix-2 and 19.03% for split-radix in the transistor count compared to the previous work.

Performance Analysis of Magnetic Resonance Image Denoising Using Contourlet Transform

A medical image denoising algorithm using contourlet transform is proposed and the performance of the proposed method is analysed with the existing methods. Noise in magnetic resonance imaging has a Rician distribution and unlike AWGN noise, Rician noise is signal dependent. Separating signal from Rician noise is a tedious task. The proposed approaches were compared with other transform methods such as wavelet thresholding and block DCT. Hard, soft and semi-soft thresholding techniques are described and applied to test images with threshold estimators like universal threshold. The results are compared based on the parameters: PSNR and MSE. Numerical results show that the contour let transform can obtained higher PSNR than wavelet based and block DCT based denoising algorithms.

Drowsiness Detection Using Photoplethysmography Signal

This study presents an innovative approach to detect drowsiness by using photoplethysmography signals which is easily acquirable with non-invasive techniques. Drowsiness detection based on biological signals is being employed in precautionary personal safety. Autonomous Nervous System (ANS) activity can be measured non-invasively from the Pulse Rate Variability (PRV) signal obtained from photoplethysmography signal (PPG), that comprises alterations during, relaxation, extreme fatigue and drowsiness episodes. Our hypothesis is that these variations manifest on PRV. In this work we develop an on-line detector of drowsiness based on PRV analysis. The databases have been collected with the aid of an external observer who decides upon each minute of the recordings as drowsy or awake, and constitutes our data base.

A novel method for glaucoma detection using optic disc and cup segmentation in digital retinal fundus images

Retinal fundus photographs has always remained the gold standard for evaluating the changes in retina. Here, a novel method for automatic glaucoma detection from digital retinal fundus images is proposed. The methodology makes use of optic disc and cup segmentation. Optic disc is segmented using morphological operations and hybrid level-set methodology. Optic cup is segmented by first detecting blood vessels using SVM classifier and then the bending points on the circum linear vessels. Parameters such as vertical cup-to-disc ratio (CDR), cup-to-disc area ratio are calculated and used for glaucoma detection. A CDR value greater than 0.5 and cup-to-disc area ratio greater than 0.3 indicates the presence of glaucoma. The proposed method is found to produce a mean error as low as 0.021 (CDR) when compared with expert observation.

Detection and localization of texts from natural scene images using scale space and morphological operations

Detection and localization of texts from natural scene images is important and can provide a much truer form of content-based image analysis if it can be extracted and harnessed efficiently. This problem becomes challenging because of complex background, variations of text font, size and line orientation, non-uniform illumination. A new unsupervised text detection algorithm is proposed in this paper. In this approach scale space and morphological operations for the edge detection are utilized. The non-text components are efficiently filter out by using scale decomposition and 2D Gaussian low pass filter. They are extracted based on observation that the edge of a character can be extracted from the complex scenes by taking into consideration the high similarities in length and aspect ratio. The proposed method yielded high precision when experiments where evaluated in the ICDAR 2003 dataset.

A novel approach for contrast enhancement and noise removal of medical images

This paper presents the application of two different image enhancement techniques to medical images and the comparison of these techniques with traditional Histogram Equalization (HE) method. The proposed method uses Weighted Histogram Equalization (WHE) and transform domain approach to enhance medical images. Simulation results shows that Perona-Malik filter (PM filter) can be used to remove the noise from the enhanced image without impacting the image contrast. Peak Signal to Noise Ratio (PSNR) values shows that the use of Perona-Malik filter can improve both the methods. The proposed WHE method with PM filter shows better PSNR values than the transform domain approach with PM filter.

Medical image enhancement by applying averaging method in clusters

In this paper a new technique to enhance the medical images without distorting the local information is proposed. Proposed method is capable of reducing the over-enhancement problem. The method first cluster the gray levels based on certain criteria and then the new transformation function is applied to each cluster. Proposed method uses Averaging method to transform the gray levels in each cluster. PSNR values show that proposed method performs better than other spatial and transform based image enhancement techniques. Experimental results show that proposed method produces noise free contrast enhanced images.

Comparison of denoising methods in diffusion tensor imaging

In this paper, a non linear adaptive Gaussian de-noising method for diffusion tensor imaging (DTI) is proposed. DTI image are of poor SNR and low resolution images. In order to improve DTI, the proposed method is applied to the diffusion weighted images (DWI) from which DTI is computed. The anisotropic flow principle is used in non linear adaptive Gaussian denoising method and smoothing will vary according to the anisotropic flow. The proposed method is compared with the scalar Partial Differential Equation (PDE) denoising method. The non linear adaptive Gaussian denoising method shows better performance compared to scalar PDE. To evaluate the efficiency of both denoising methods, image quality metrics like peak signal-to-noise ratio (PSNR) and mean structural similarity index measure (MSSIM) are used. The experimental results indicate the good performance of proposed method and it has a better denoising effect in DTI compared to scalar PDE.

Contrast enhancement based denoising method in diffusion tensor imaging

In this paper, a new denoising and contrast enhancement method for DTI is proposed. Noise removal is given priority in existing denoising methods. In order to increase the visibility of structural details, contrast enhancement methods has to be used. In proposed method, a non-linear adaptive Gaussian denoising filter removes noises from the DTI. To increase the visibility of filtered micro structural details, a contrast enhancement method is also used. The proposed method is compared with the existing scalar Partial Differential Equation (PDE) and non local means (NLM) denoising methods. Quantitative measures are used for validating the efficiency of the proposed contrast enhancement method. The experiment results shows that the proposed method outperforms the scalar PDE method and NLM method.

Reconstruction of Beat Signal in Radar Altimeter Using Signal Processing

The frequency modulated continuous wave (FMCW) radar principle has been used in altimeter to measure altitude above the surface of the Earth. Traditionally it has been used in short range application due to its unambiguous range. The enhanced range resolution is a factor for FMCW radars compared with other types of radars. There are theoretical restrictions in the range resolution. This paper deals with reconstruction of beat signal in the discontinuous region of its samples. Furthermore, the signal processing methods like FFT calculations that are used for FMCW radar signals and the possible improvement techniques for these methods are discussed. This technique overcomes the discontinuities in the turnaround region of radar altimeter beat signal for improved FFT measurement. The irregularities are due to the delay of received signal in that region and to overcome this problem, we take the similar samples from both sides of the turnaround region and reconstruct the original signal.