The Field Programmable Gate Array (FPGA) technology is widely used in various applications such as Digital imageprocessing (DIP), Digital signalprocessing (DSP), data processing, etc. As the active and the passive c...
详细信息
ISBN:
(纸本)9781467393393
The Field Programmable Gate Array (FPGA) technology is widely used in various applications such as Digital imageprocessing (DIP), Digital signalprocessing (DSP), data processing, etc. As the active and the passive components are fabricated in a single chip the performance level of the applications, the overall speed, area, and power of the application systems can be minimized effectively. The identification of the circuit information is an important process in the very Large Scale Integration (vLSI) design. The existing architectures such as neural network, and brute force approach, etc., improve the analytical model work and refine the model using the trial and error method. The use of these existing techniques for the circuit information identification is inefficient, inaccurate, highly complex, time consuming, power consuming, and produced minimal classification result quality. To address these issues we have proposed a FPGA based vLSI architecture. The proposed architecture collects the feature values of the circuit manually. The collected features are then applied the Discrete Fourier Transform (DFT) based feature extraction process for identifying the information for the required data base and the input selected circuit details. The classification process is performed using the Knowledge Based Neural Network (KBNN). The performance of the proposed architecture is compared with the existing neural network methodology. The comparison results show that the suggested architecture increases the speed, reduces the overall power consumption, overall delay time, and also improves the circuit details classification speed level.
A wide class of nonlinear wavelet-like transforms (NLWT) is introduced. Inside it, a subclass of NLWT is built with the structure of a fast algorithm. Each fast transformation from this class is represented as a paral...
A wide class of nonlinear wavelet-like transforms (NLWT) is introduced. Inside it, a subclass of NLWT is built with the structure of a fast algorithm. Each fast transformation from this class is represented as a parallel-serial superposition of nonlinear (2x2)-transforms. The reported architecture generalizes both linear and non-linear fast transforms, which can be considered as a formal framework for generalized signalprocessing. This approach leads to a number of new nonlinear transforms of potential interest for imageprocessing.
Current source inverter(CSI)fed drives are employed in high power *** conventional CSI drives suffer from drawbacks such as harmonic resonance,unstable operation at low speed ranges,and torque *** fed drives with Dire...
详细信息
Current source inverter(CSI)fed drives are employed in high power *** conventional CSI drives suffer from drawbacks such as harmonic resonance,unstable operation at low speed ranges,and torque *** fed drives with Direct Torque Control(DTC)has drawn the attention of the motor drives designers because its implementation requires no position *** to the success of this scheme is the estimation of electromagnetic torque and stator flux linkages using the measured stator voltages and *** estimation is dependent only on one machine parameter,stator *** variation of the stator resistance,deteriorates the performance of the drive by introducing errors in the estimated flux linkage's magnitude and its position and hence in the electromagnetic *** change also skews the torque linearity thus making the motor drive a less than ideal torque *** compensation using stator current phasor error has been proposed in *** obtain the stator current phasor error,the stator current reference is required which is not usually available in direct torque control *** analytical derivation of the stator current phasor reference is derived systematically from the reference electromagnetic torque and flux *** error between the stator current phasor reference and its measured value is a measure of the stator resistance variation from its set *** the first time,it is demonstrated in this paper that DTC motor drive system can become unstable when the set value of the stator resistance in the controller is higher than the stator resistance in the *** parameter adaptation is not only important for torque linearity but also for stability of the system is shown in this paper.
Quaternion wavelet transform (QWT) combines discrete wavelet transform (DWT) and quaternion Fourier transform (QFT). QWT has many applications included imageprocessing. In this research, we discuss about construction...
Quaternion wavelet transform (QWT) combines discrete wavelet transform (DWT) and quaternion Fourier transform (QFT). QWT has many applications included imageprocessing. In this research, we discuss about construction, characteristics and implementation of QWT on process of image denoising. We construct denoising algorithm with QWT then we do simulation to know performance of algorithm. We use grayscale test images that have size 512 × 512 pixel with low, medium and high complexity. Experiment removes noise of image successfully. Results of image denoising are used to measure algorithm performance using PSNR (peak signal to noise ratio) value. We compare PSNR values with DWT and QWT for Haar, Biorthogonal, Daubechies and Coiflets wavelet. The method that has the highest PSNR value can be concluded the best performance.
Brain computer interface (BCI) is a widely used system to assist the disabled and paralyzed people by creating a new communication channel. Among the various methods used in BCI area, motor imagery (MI) is the most po...
详细信息
ISBN:
(纸本)9781467311496
Brain computer interface (BCI) is a widely used system to assist the disabled and paralyzed people by creating a new communication channel. Among the various methods used in BCI area, motor imagery (MI) is the most popular and the most common one due to its the most natural way of communication for the subject. Some software applications are used to implement BCI systems, and some toolboxes exist for EEG signalprocessing. In recent years virtual reality (vR) technology has entered into the BCI research area to simulate the real world situations and enhance the subject performance. In this work, a completely MATLAB-based Mi-based BCI system is proposed and implemented in order to navigate into a virtual environment. In addition, a variety of features types were employed to select the best ones in the proposed system with the use of linear discriminant analysis (LDA) classifier through some interactive graphical user interfaces (GUIs). The results show the feasibility of the proposed BCI system in the subject training with or without feedback and even navigation into a virtual home.
Convolutional Neural Networks (CNN's) are known to perform well on computer vision tasks such as image classification, image segmentation, and object detection. However, one major drawback of CNN's is the huge...
Convolutional Neural Networks (CNN's) are known to perform well on computer vision tasks such as image classification, image segmentation, and object detection. However, one major drawback of CNN's is the huge amount of computing and memory resources needed to train them. In this paper, we propose an architectural unit which we call Upsampling-Based wavelet Residual Block (UBWRB), that utilizes the 2D discrete wavelet transform coupled with upsampling operators and a residual connection to extract features from image data while having relatively fewer trainable parameters as compared to traditional convolutional layers. The discrete wavelet transform is a family of transforms that find extensive applications in signalprocessing and time-frequency analysis. For this paper, we use the filter-bank implementation of the discrete wavelet transform, allowing it to act in a similar fashion to a convolutional layer with fixed kernel weights. We demonstrate the performance and parameter-efficiency of CNN's with UBWRB's in the task of image classification by training them on the MNIST, Fashion-MNIST, and CIFAR-10 datasets. Our best-performing models achieve a test accuracy of 99.34% on the MNIST dataset while having less than 120,000 trainable parameters, and 92.90% and 84.27% on the Fashion-MNIST and CIFAR-10 datasets respectively, with both having less than 180,000 trainable parameters.
The article presents the results of studies on interference suppression in optoelectronic methods for monitoring weft thread weaving looms on the basis of linear image sensors caused by inhomogeneities of the backgrou...
The article presents the results of studies on interference suppression in optoelectronic methods for monitoring weft thread weaving looms on the basis of linear image sensors caused by inhomogeneities of the background created by sources of natural or artificial light sources in a controlled image scene. Existing solutions that use linear or nonlinear filtering are focused primarily on the processing of two-dimensional images on personal computers and are unsuitable for use in embedded applications. In particular, their use is also ineffective for solving the above problem, when the useful signal is defined as the difference between the output signals of the photodetector taken at a fixed time interval. As studies have shown, in this case, the best result is the use of discrete wavelet-transform.
Deaf people all around the world use sign language to communicate and like oral languages vary from country to another so it is for the sign languages. In this paper, we propose a probabilistic neural network (PNN) fo...
详细信息
ISBN:
(纸本)9781479982134
Deaf people all around the world use sign language to communicate and like oral languages vary from country to another so it is for the sign languages. In this paper, we propose a probabilistic neural network (PNN) for two Sign languages: American Sign Language (ASL) recognition for static signs and Arabic sign Language. The signs in both of them are realized with one naked hand and simple background. DCT, DWT and PCA for spatial reduction method. Although PCA has been used before in sign language as a dimensionality reduction technique, it is used here as a descriptor that represents a global image feature. Finally we combine the features to improve the recognition rate (RR) and an error rate (ER) where DWT combined with the PCA using PNN classifier achieves RR 80.2% and ER 3.90% for Arabic database. The RR is improved to be 94% for American database with an ER 1.2%.
Optical coherent tomography (OCT) is a rapidly developing method of fundamental and applied research. Detection and processing of OCT images is a very important problem of applied physics and optical signalprocessing...
Optical coherent tomography (OCT) is a rapidly developing method of fundamental and applied research. Detection and processing of OCT images is a very important problem of applied physics and optical signalprocessing. In the present paper we are demonstrating the ability for effective wavelet-domain de-noising of OCT images. We are realizing an algorithm for wavelet-domain de-noising of OCT data and implementing it for the purpose of studying test samples and for in vivo nail tomography. High de-noising efficiency with no significant losses of information about the internal sample structure is observed.
The Quality of vision is a practical objective in the image *** of the applications of imageprocessing procedures is image Fusion. To get the subjective vision of a image by gathering the best data from source images...
The Quality of vision is a practical objective in the image *** of the applications of imageprocessing procedures is image Fusion. To get the subjective vision of a image by gathering the best data from source images of a similar scene/picture and spot them in a solitary void image is the image Fusion process. A simple and versatile approaches are statistical measures, that are applicable in any kind of image/signalprocessing techniques. In the first step, image is decomposed using Discrete wavelet Transform (DWT). Mathematical computation of Smoothness is proposed in transform domain. This metric is used to select important information from multiple source images resulting to fused image. Further, Human visual System (HvS) is also explored for fusion. Here all sub bands of DWT are multiplied with HvS weights. Highest response sub band is identified from various sub bands of multiple images using HvS. These sub bands are selected to get the fused image. Smoothness based fusion technique identifies the good texture information and leave the noise affected portions from the multiple source images. HvS based fusion technique identifies the visually important information from the multiple source images. The registered multi-focus and medical images are considered as source images. The experimental results shows that proposed fusion techniques are good in terms of popular fusion metrics.
暂无评论