Vishnu V. Makkapati

Encoding of multispectral and hyperspectral image data using wavelet transform and gain shape vector quantization

An effective and lossy compression technique for multispectral and hyperspectral image data minimizes both the spatial and spectral correlations while preserving the spectral characteristics of the data. In this paper, we combine wavelet transform and a variant of vector quantization for decorrelating both spatial and spectral information, and thus aim to achieve superior quality. We use 2-D wavelet transform followed by Kronecker-Product Gain-Shape Vector Quantization. This is coupled with the generalized BFOS for obtaining an optimal bit-rate. The pixels within the subbands of multi- and hyper-spectral images exhibit greater spectral redundancy which is thus exploited by designing multiresolution codebooks. Results are presented for multispectral and hyperspectral data taken from different sensors in different bands. The results obtained with our scheme are compared with other techniques designed for multi-/hyper-spectral image data and the Wavelet-based JPEG-2000. The computational requirement of the proposed technique is lower in comparison with other vector-quantization techniques.

Performance enhancement of a spline-based method for extreme compression of weather radar reflectivity data

Enhancements are carried out to a contour-based method for extreme compression of weather radar reflectivity data for efficient storage and transmission over low-bandwidth data links. In particular, a new method of systematically adjusting the control points to obtain better reconstruction of the contours using B-spline interpolation is presented. Further, bit-level manipulations to achieve higher compression ratios are investigated. The efficacy of these enhancements is quantitatively evaluated with respect to achievable compression ratios and the root-mean-square error of the retrieved contours, and qualitatively assessed from the visual fidelity of the reconstructed contours. Results are presented for actual Doppler weather radar data sets

Synthesis of generalized vertical-plane weather radar imagery along aircraft flight paths

A method for synthesizing weather pictures in vertical planes along aircraft flight paths is presented. The weather data are derived from a number of Doppler radars covering different parts of the flight path. The flight path consists of straight segments with arbitrary offset and orientation with respect to individual radar locations. The intersection of radar scanning cones with the vertical plane segments are described as families of radial lines or hyperbolas depending on whether the flight path passes over a given radar or is offset from it. Radar data may be stretched or interpolated between radials to achieve continuity and smoothness of display. Schemes are incorporated for optimally using redundant data from multiple radars. Both scalar and vector weather data may be displayed. Such display is of great value for flight planning.

Contour Encoding Based on Extraction of Key Points Using Wavelet Transform

In many situations, it is convenient to represent pictorial data in the form of contours. It may become necessary to compress such contour data for efficient storage and transmission. We present here a technique for achieving very high levels of compression of 2D contours. The goal here is to represent each contour using a discrete set of representative points known as key points. A novel method of extracting the key points using wavelet transform is presented. The scheme exploits the properties of the high frequency coefficients to identify these points. Local peaks in the magnitude plot of high frequency coefficients are designated as key points and are identified using an efficient algorithm. The performance of the scheme is evaluated using multiple actual contours derived from weather radar reflectivity fields

Camera based respiration rate of neonates by modeling movement of chest and abdomen region

Respiration Rate (RR) is one of the important parameters used for monitoring of neonates in a Neonatal Intensive Care Unit (NICU). In this paper, we propose a contactless method of measuring RR using a camera. Our algorithm exploits the motion of the chest and abdomen during the breathing cycle of a neonate. The chest and abdomen regions will typically expand and contract during inspiration and expiration phases respectively. We model this natural movement by capturing the motion of these regions using optical flow. The optical flow vectors converge and diverge during expiration and inspiration respectively and we capture this pattern by using a convergence and divergence filter. We use Independent Component Analysis (ICA) to separate the natural motion of the chest and abdomen from noise. To reduce the computational complexity, we use the convergence and divergence filter in Clifford Fourier Transform (CFT) of the vector field. We show the efficacy of our method by comparing the results with the ground truth RR collected from a patient monitor.

Vector Quantization Using Reflections of Triangular Subcodevectors

The design of a codebook consisting of codevectors is the goal of vector quantization (VQ) schemes. This solution is not unique and many variants of VQ have been proposed. VQ suffers from computational and memory complexities that increase with the size of codebook and codevector dimensions. The lack of a universal codebook that works across different class of images necessitates the transmission of the codebook along with the codevector indices. This paper proposes a novel method of vector quantization using reflections of triangular subcodevectors. This method jointly reduces the memory and computational resources required for VQ. Numerical results are presented for the proposed scheme in comparison to a traditional VQ method and a method based on reflections

Studies on a High-Compression Technique for Weather Radar Reflectivity Data

Studies are carried out on a contour-based method for highly compressing voluminous weather radar reflectivity data to optimize the compression and reconstruction parameters. In particular, the effects of the choice of thresholds and the degree of smoothing at encoding and retrieval ends are studied with respect to achievable compression ratios, root-mean-square error of the retrieved contours, and the visual fidelity of the reconstructed reflectivity images. Results are presented for multiple actual Doppler weather radar reflectivity data sets

Swapping of reflections of adjacent subcodevectors for vector quantization

Vector quantization (VQ) has been widely used for compressing images. VQ of images requires significant memory and computational resources. Many methods have emerged to reduce either of these resources. A class of methods aimed at reducing the memory utilizes reflection of codevectors. This paper analyzes a scheme based on swapping of reflections of adjacent subcodevectors for jointly reducing memory and computational overheads. This scheme extracts the intra-vector redundancies to provide optimal quality. Studies are carried out on the convergence, utilization of the orientations, and performance of the scheme. Results are presented in comparison to a traditional scheme based on reflections of codevectors.