Mohsin M. Jamali

A VLSI architecture for real-time image coding using a vector quantization based algorithm

Digital image coding using vector quantization (VQ) based techniques provides low-bit rates and high quality coded images, at the expense of intensive computational demands. The computational requirement due to the encoding search process, had hindered application of VQ to real-time high-quality coding of color TV images. Reduction of the encoding search complexity through partitioning of a large codebook into the on-chip memories of a concurrent VLSI chip set is proposed. A real-time vector quantizer architecture for encoding color images is developed. The architecture maps the mean/quantized residual vector quantizer (MQRVQ) (an extension of mean/residual VQ) onto a VLSI/LSI chip set. The MQRVQ contributes to the feasibility of the VLSI architecture through the use of a simple multiplication free distortion measure and reduction of the required memory per code vector. Running at a clock rate of 25 MHz the proposed hardware implementation of this architecture is capable of real-time processing of 480*768 pixels per frame with a refreshing rate of 30 frames/s. The result is a real-time high-quality composite color image coder operating at a fixed rate of 1.12 b per pixel. >

The application of Evolutionary Neural Network for bat echolocation calls recognition

An Evolutionary Neural Network (ENN) is developed to identify bats by their vocalization characteristics. This is in an effort to identify local bat species as a large number of bat fatalities near wind turbines have been reported. ENN is based on the Genetic Algorithm, which can be used for optimization of the weight selection of the neural network. We then compare ENN with different classification techniques. In the scope of bat call classification, ENN is a new technique that can be effectively used as a bat-call classifier. This research will help in developing mitigation techniques for reducing bat fatalities. The ENN algorithm is developed in MATLAB.

Avian detection & tracking algorithm using infrared imaging

This paper presents a method for target detection and tracking of IR images in the application of avian surveillance. As there are many reports of avian mortality due to collision with turbine blades, the detection and tracking of birds at turbine sites is an important issue. In this work, three different background subtraction techniques are first applied to detect moving objects. Otsu thresholding method is then extended by incorporating an adaptive variable based on the mean of each frame and certain constant value. Filtering using morphological operations is applied. Results of three different techniques are then compared. Selected technique (RA) followed by thresholding and filtering is then used for tracking and information extraction. Results show that proposed method provides the needed accuracy for IR imagery. This method can be effectively used in different applications of IR imaging.

Implementation of ant clustering algorithm for IR imagery in wind turbine applications

Interaction of avian with turbines has become an important public policy issue, so identification and quantification of avian at turbine sites is crucial.

A novel feature extraction algorithm for classification of bird flight calls

Acoustic monitoring of birds in the vicinity of wind turbines is becoming an important public policy issue. Acoustic monitoring involves preprocessing, feature extraction and classification. A novel Spectrogram-based Image Frequency Statistics (SIFS) feature extraction algorithm has been developed. Features extracted from proposed algorithms were then combined with various classification algorithms such as k-NN, Multilayer Perceptron (MLP) and Hidden Markov Models (HMM) and Evolutionary Neural Network (ENN). SIFS and MMS algorithms, combined with ENN, provided the most accurate results. Proposed algorithms were tested with real data collected during spring migration around Lake Erie in Ohio.

Development of a FPGA-based high speed FFT processor for wideband Direction of Arrival applications

A parallel and pipelined Fast Fourier Transform (FFT) processor for use in the Direction of Arrival (DOA) estimation of a wideband waveform is presented. The selected DOA algorithm follows the Coherent Signal Subspace Method (CSSM). The target device for implementation is a Xilinx Virtex-5 Field Programmable Gate Array (FPGA). The FFT processor was developed in MATLAB Simulink using the Xilinx System Generator block-set to auto-generate VHDL code. Although the parallel and pipelined architecture uses a large portion of the available FPGA resources, the architecture does yield a high throughput.

Optimized FPGA based implementation of particle filter for tracking applications

Particle filter has been proven to be a very effective method for identifying targets in non-linear and non-Gaussian environment. However, particle filter is computationally intensive. So, particle filter has been implemented on FPGA by exploiting parallel and pipelining approaches to reduce the computational burden. Our optimized FPGA implementation improves up to twelve times speed up. Also more speed ups are achieved with increasing number of particles.

Data Fusion of Acoustics, Infrared, and Marine Radar for Avian Study

A multisensor data fusion approach via acoustics, infrared camera, and marine radar is proposed and described in the application of avian monitoring. The ultimate scope of avian monitoring is to preserve the population of birds and bats, especially those listed in the endangered list, by observing their activity and behavior over the migration period. With the significant attention toward the construction of off-shore/on-shore wind farms in recent decades, the wind turbines are more threatening the avian life with increasing the risk of birds/bats collision with turbine blades. In order to address this problem, this paper proposes a fuzzy Bayesian-based multisensory data fusion approach to provide the activity information regarding the targets in the application of avian monitoring. The developed technique is used to process the Spring and Fall 2011 migration period.

Optimize hardware with fixed-point variable length phase factors

Fast Fourier Transforms (FFTs) are highly parallel in nature and consist of simple addition, subtraction, and complex rotation operators with phase factors (a.k.a. twiddle factors). With the advent of FPGAs and other reconfigurable seas-of-logic, it is now possible to construct a fully parallel FFT structure where the phase factors are now constants and good targets for hardware optimization. By varying the fixed-point length of the phase factors using phase angle error percentage as a control for the variable length phase factor quantizer, the number of shifted adders required to implement the complex rotation operators can be reduced. Performance comparisons of fixed length and variable length phase factors, along with two quantizer rounding modes, are investigated.

Real-time VLSI architecture for a VQ-based high-quality image coding algorithm

A real-time vector quantizer (VQ) architecture for broadcast quality encoding of color TV images is presented. The architecture maps the mean/quantized residual vector quantizer (MQRVQ), an extension of mean/residual VQ, onto a VLSI/LSI chip set. The MQRVQ contributes to the feasibility of the VLSI architecture through the use of a simple multiplication-free distortion measure and reduction of the required memory per codevector. In other words, the complexity-reduced MQRVQ allows each subcodebook of 128 codevectors (10 kb) and its associated encoding elements to be fitted onto a VLSI chip. This architecture reduces encoding search complexity through partitioning of a large codebook into on-chip memories of a concurrent VLSI chip set. There are 64 of these VLSI chips to accommodate a codebook with size up to 2/sup 13/=8192 codevectors. The architecture is capable of real-time processing of 480*768 pixels per frame with refreshing rate of 30 frames/second.<<ETX>>