We present a platform that will aid researchers in developing algorithms for specific (primarily image processing) tasks by providing training data sets with ground truth and with providing evaluation of the outputs o...
详细信息
ISBN:
(纸本)9781479972807
We present a platform that will aid researchers in developing algorithms for specific (primarily image processing) tasks by providing training data sets with ground truth and with providing evaluation of the outputs of the algorithms in an objective manner under identical conditions using standardized measures.
Most of the traditional document image classification techniques concentrate on document segmentation and OCR analysis, in spite of so many complexities and limitations involved. Recently, many of the document image c...
详细信息
ISBN:
(纸本)9781479918058
Most of the traditional document image classification techniques concentrate on document segmentation and OCR analysis, in spite of so many complexities and limitations involved. Recently, many of the document image classification problems are easily solved just by adapting standard computer vision approaches for natural image retrieval and classification, that are referred as visual appearance based document classification techniques. These approaches have reported better results as compared to the traditional approaches on proprietary datasets. However, so far these approaches are not compared with each other and, despite having potential, they are not evaluated on distorted camera-captured documents, which is one of the challenging requirements in our present commercial document analysis projects. In this paper, we present simple and effective descriptions of different visual appearance based document image classification techniques. We compare their performance on various standard and publicly available datasets, that are differ in degree of image degradations and content variations. We also demonstrate their advantages and limitations. Additionally, we make the implemented versions of these method publicly available to research community for usage and further testing on other domains.
Spectrum awareness is one of the most challenging tasks in cognitive radio (CR). To adequately adapt to the changing radio environment, it is necessary for the CR to be able to perform joint detection and classificati...
详细信息
ISBN:
(纸本)9781424497218
Spectrum awareness is one of the most challenging tasks in cognitive radio (CR). To adequately adapt to the changing radio environment, it is necessary for the CR to be able to perform joint detection and classification of low signal-to-noise ratio (SNR) signals without requiring much a priori information on the signal parameters. In this paper, we propose a joint detection and classification algorithm for the mobile Worldwide Interoperability for Microwave Access (WiMAX) and Long Term Evolution (LTE) signals. The algorithm has the advantage that it requires relaxed information on the signal parameters. Simulation results are presented, which show the efficiency of the proposed algorithm under diverse scenarios.
Parkinson's disease (PD) is a chronic neurological progressive disorder caused by lack of the chemical dopamine in the brain. Up to today, there is still no cure or prevention for PD, and usually the disease worse...
详细信息
ISBN:
(纸本)9781479979738
Parkinson's disease (PD) is a chronic neurological progressive disorder caused by lack of the chemical dopamine in the brain. Up to today, there is still no cure or prevention for PD, and usually the disease worsens gradually over time. However, this disease can be controlled with some treatment, especially in the early stage. Hence, this study proposes a method in early detection and diagnosis of PD by using the Multilayer Feedforward Neural Network (MLFNN) with Back-propagation (BP) algorithm. This MLFNN with BP algorithm is simulated using MATLAB software. The dataset information used in this study was taken from the Oxford Parkinson's Disease Detection Dataset. The output of the network is classified into healthy or PD by using K-Means Clustering algorithm. The performance of this classifier was evaluated based on the three parameters;sensitivity, specificity and accuracy. The result shows that network can be used in diagnosis and detection of PD due to the good performance, which is 83.3% for sensitivity, 63.6% for specificity, and 80% for accuracy.
Research and industry increasingly rely on image processing to analyze an environment. Most image processing requires significant computing power and consequently a complex processing unit. This paper presents a hardw...
详细信息
ISBN:
(纸本)9781479947744
Research and industry increasingly rely on image processing to analyze an environment. Most image processing requires significant computing power and consequently a complex processing unit. This paper presents a hardware specific platform that computes common image processing algorithms and extracts the information from a video stream. By this arrangement, the image processing can be customized for a specific application, using the same embedded system. This platform also simplifies the development of image processing systems and algorithms by focusing on higher level algorithmic operations.
In the framework of the Urban Atlas 2012 production, this paper investigated a set of generative models (Maximum likelihood, k-means) and discriminative models (k Nearest Neighbors, Support Vector Machine and Neural N...
详细信息
ISBN:
(纸本)9781479966523
In the framework of the Urban Atlas 2012 production, this paper investigated a set of generative models (Maximum likelihood, k-means) and discriminative models (k Nearest Neighbors, Support Vector Machine and Neural Network) to extract urban-tree cover at a European scale. Based on SPOT-5 images and a training on a large coarse resolution dataset, this study tested the performance of these algorithms on 3 cities regarding their geographical location, urban morphology and acquisition dates. Result reveals that discriminative models are more robust than generative ones. It shows that overall accuracy varies from 75% for the k-means classifier to 85% for the neural network. It also shows that neural networks provide the most balanced results (ratio between commission and omission errors) leading to be most suitable algorithm to process different cities with heterogeneous data.
Despite recent advancements, the time, skill, and monetary investment necessary for hardware setup and calibration are still major prohibitive factors in field data sensing. The presented research is an effort to alle...
详细信息
ISBN:
(纸本)9781479974863
Despite recent advancements, the time, skill, and monetary investment necessary for hardware setup and calibration are still major prohibitive factors in field data sensing. The presented research is an effort to alleviate this problem by exploring whether built-in mobile sensors such as global positioning system (GPS), accelerometer, and gyroscope can be used as ubiquitous data collection and transmission nodes to extract activity durations for construction simulation input modeling. Collected sensory data are classified using machine learning algorithms for detecting various construction equipment actions. The ability of the designed methodology in correctly detecting and classifying equipment actions was validated using sensory data collected from a front-end loader. Ultimately, the developed algorithms can supplement conventional simulation input modeling by providing knowledge such as activity durations and precedence, and site layout. The resulting data-driven simulations will be more reliable and can improve the quality and timeliness of operational decisions.
In this paper, we address the problem of state estimation of linear switched discrete time models from a finite set of input-output data. This is a challenging problem since it requires estimating the active discrete ...
详细信息
ISBN:
(纸本)9781479917587
In this paper, we address the problem of state estimation of linear switched discrete time models from a finite set of input-output data. This is a challenging problem since it requires estimating the active discrete state and its continuous observer. In fact, we propose a hybrid state observer design which consists of two stages. The first allows to determine the active discrete state using a classification algorithm that associated the current data to its appropriate submodel. The second stage is used to estimate the corresponding continuous observer. Simulation results are presented to illustrate the performance of the proposed method.
Modular multilevel converter (MMC) has become a promising technology for medium/high voltage high power applications. Electric vehicles on the other hand, have a large number of series connected battery cells to enhan...
详细信息
ISBN:
(纸本)9781509066841
Modular multilevel converter (MMC) has become a promising technology for medium/high voltage high power applications. Electric vehicles on the other hand, have a large number of series connected battery cells to enhance the output voltage to drive the connected motor. This paper applies the concept of modular multilevel converter for low power electric vehicle applications where, the batteries of the EV help in exchanging power with the motor. The batteries can discharge when feeding power to the motor and get charged when the motor runs into regenerative mode. Multi-carrier PWM technique is employed to generate the multilevel voltage waveforms and sorting technique is used to balance the battery voltages. Simulation study has been done in MATLAB/SIMULINK platform and it is observed that the voltages and State-of-Charge (SOC) of batteries are well maintained.
In this paper, fault detection in HP drum of boilers in Kerman combined cycle power plant is explored by means of support vector machine (SVM) algorithm and principal component analysis (PCA). Initially, SVM classifie...
详细信息
ISBN:
(纸本)9781479931170
In this paper, fault detection in HP drum of boilers in Kerman combined cycle power plant is explored by means of support vector machine (SVM) algorithm and principal component analysis (PCA). Initially, SVM classifier algorithm and PCA are discussed and then based on the collecting data on normal and abnormal operating the conditions of boilers, fault detection is carried out via explained methods. Finally, a comparison of these techniques and other routine methods is made to show the superiority with the proposed approaches in Kerman power plant.
暂无评论