Showing papers by "Andreas Spanias published in 2008"

Shift-invariant sparse representation of images using learned dictionaries

Abstract: Sparse approximations that are evaluated using over complete learned dictionaries are useful in many image processing applications such as compression, denoising and feature extraction. Incorporating shift invariance into sparse representation of images can improve sparsity while providing a good approximation. The K-SVD algorithm adapts the dictionary based on a set of training examples, without shift invariance constraints. This paper presents two algorithms for training dictionaries and evaluating shift-invariant sparse representations for image data. One is a modified version of the K-SVD algorithm and the other is a novel graph-based algorithm that adapts the dictionary and computes representations using a low complexity reconstruction procedure.

Fast query by example of environmental sounds via robust and efficient cluster-based indexing

Abstract: There has been much recent progress in the technical infrastructure necessary to continuously characterize and archive all sounds, or more precisely auditory streams, that occur within a given space or human life. Efficient and intuitive access, however, remains a considerable challenge. In specifically musical domains, i.e., melody retrieval, query-by-example (QBE) has found considerable success in accessing music that matches a specific query. We propose an extension of the QBE paradigm to the broad class of natural and environmental sounds, which occur frequently in continuous recordings. We explore several cluster-based indexing approaches, namely non-negative matrix factorization (NMF) and spectral clustering to efficiently organize and quickly retrieve archived audio using the QBE paradigm. Experiments on a test database compare the performance of the different clustering algorithms in terms of recall, precision, and computational complexity. Initial results indicate significant improvements over both exhaustive search schemes and traditional K- means clustering, and excellent overall performance in the example-based retrieval of environmental sounds.

Real-time sensing and acoustic scene characterization for security applications

Abstract: Real-time acoustic scene analysis has several applications such as homeland security, surveillance, and monitoring. The development of a collaborative networking infrastructure can be valuable in scene analysis since feature parameters can be extracted locally (at the node level) and combined at the base station. In this context, distributed and agile wireless sensor networks (WSNs) have been of particular interest recently. In this paper, we propose real-time voice scene characterization algorithms for use in a wireless sensor network. Voice scene analysis is accomplished using a speech discriminator, a gender classifier, a system for recognizing the state of emotion, and an estimator of the number of speakers in an area of interest. Real-time implementations of these algorithms are accomplished using Crossbow motes and TI DSP boards, configured to operate in a wireless sensor network. A series of experiments are presented that characterize the performance of the algorithms under different conditions.

Sparse Representations for Pattern Classification using Learned Dictionaries

Abstract: Sparse representations have been often used for inverse problems in signal and image processing Furthermore, frameworks for signal classification using sparse and overcomplete representations have been developed Data-dependent representations using learned dictionaries have been significant in applications such as feature extraction and denoising In this paper, our goal is to perform pattern classification in a domain referred to as the data representation domain, where data from different classes are sparsely represented using an overcomplete dictionary We propose a source model to characterize the data in each class and present an algorithm to infer the dictionary from the training data of all the classes We estimate statistical templates in the data representation domain for each class of data, and perform classification using a likelihood measure Simulation results show that, in the case of highly sparse signals, the proposed classifier provides a consistently good performance even under noisy conditions

A low-complexity loudness estimation algorithm

Abstract: Audio processing applications such as rate determination, bandwidth extension, compression, and noise reduction make use of loudness metrics. Most loudness estimation algorithms are computationally expensive and often not suitable for real time applications. In this paper, we present a low-complexity loudness estimation algorithm applicable to both steady and time-varying sounds. The model computes an estimate of the excitation pattern by simultaneously pruning the frequency components and detector locations. Comparative results indicate that the proposed algorithm performs consistently well for different types of audio signals at a reduced complexity.

Split-band speech compression based on loudness estimation

Abstract: A frame is received that has the wideband audio signal. The low band audio signal is encoded to generate an encoded low band signal. The high band signal is analyzed to determine whether the high band signal is perceptually relevant to the low band signal. If the high band signal is not perceptually relevant to the low band signal, the low band signal is encoded and provided in a frame to the decoder without including parameters corresponding to characteristics of the high band signal. If the high band signal is perceptually relevant, the high band signal is encoded to generate an encoded high band signal. The resultant frame that is sent to the decoder will include a combination of the encoded low band signal and the encoded high band signal.

Performance of distributed estimation over multiple access fading channels with partial feedback

Abstract: We consider a wireless sensor network for distributed estimation over Rayleigh fading channels. The sensors transmit their observations over fading channels to a fusion center, where a source parameter is estimated. Since the sensor transmissions add incoherently over a multiple access channel, we consider partial channel knowledge at the sensors to improve performance. We calculate the variance of the estimate when the channel phase is quantized uniformly and fed back to the sensors. We show that as few as 3 bits of feedback is sufficient for a loss in performance of about 5%. We also show that the performance is robust in the presence of feedback errors.

DNA sequence alignment using the matching pursuit decomposition

Abstract: Sequence alignment is the positioning of primary biological sequences, such as DNA, RNA and protein sequences, to identify regions of similarity in large databases. Common signal processing techniques include cross-correlations in time or frequency. However, these techniques can result in many misalignments when capturing a grouping in local or repetitive portions of the sequence. We propose a time-frequency based alignment technique using the matching pursuit decomposition method and a mapping algorithm. The aim of this alignment technique is to identify local and global alignments more efficiently and with greater precision than existing methods. Its success is based on the fact that sequence elements are mapped to unique Gaussian basis atoms that uniformly sample the time-frequency plane.

Work in progress - Java simulations of DSP algorithms for ion-channel sensors

Abstract: The use of ion channels as sensing elements for chemical and biological agents is a rapidly developing area. Ion channels are proteins that mediate the flow of ions and molecules across membranes such as cell walls. At Arizona State University researchers have devised a silicon ion-channel sensor. Experiments have been conducted to characterize this sensor and examine its utility in various applications. This paper presents Java functions developed to demonstrate ion channel signals and their analysis using DSP functions in class. The Java functions were developed in the J-DSP visual programming environment. Students can experiment with ion-channel signals, extract features, and differentiate signals representing the presence of different analytes.

Ultrasound imaging media layer texture analysis of the carotid artery

Abstract: The intima-media thickness (IMT) of the common carotid artery (CCA) is widely used as an early indicator of cardiovascular disease (CVD). It was proposed but not thoroughly investigated that the media layer (ML), its composition and texture, may be indicative for identifying the risk of stroke and differentiating between patients of high and low risk. In this study we investigate the usefulness of texture analysis of the ML of the CCA. The study was performed on 100 longitudinal ultrasound images acquired from asymptomatic subjects at risk of atherosclerosis. The images were separated into three different age groups, namely below 50, 50 to 60, and above 60 years old. A total of 61 different texture features were extracted from the intima-media complex (IMC), ML and the intima layer (IL). The IMC and ML were segmented manually by a neurovascular expert and automatically by a snakes segmentation system. It was shown that texture features extracted from the IL, ML and IMC are significantly different (mean, gray scale median (GSM), standard deviation, contrast, difference variance, periodicity) and that some of them can be associated with the increase (difference variance, entropy) or decrease (GSM) of patientpsilas age. It was also shown that the GSM of the ML falls linearly with increasing ML thickness (MLT) and with increasing age. Further research on more subjects is required for estimating other features that may provide information for patients at risk of stroke.