Search

This paper deals with the problem of automatic handwritten text detection. Paper forms are used to gather significant amount of information from people in most part of the world. While this handwritten text eventually gets digitized down the line, it takes a great amount of time and resources to do so. The objective of this paper is to reduce that amount of time significantly by introducing an easy to use system. For that purpose, we have created a web based software which takes in images and automatically recognizes the hand written elements. It then stores that information in a database which can later be evaluated and corrected by human intelligence. We have found that this kind of semi-automatic form filling technique can reduce the time it takes to correctly fill out one form. Color based image segmentation technique was used to separate the hand written text from the printed text. K-Means algorithm was applied for cluster recognition during the segmentation phase. Afterward, MATLAB's Optical Character Recognition functionality was used to extract meaningful text from the segmented image. [ABSTRACT FROM AUTHOR]

Strategic planning for all departments at all levels of government depend highly on accurate data collection. Planning and deploying resources becomes inefficient if accurate and sufficient data is not available for some geographical locations. In this research work, we intend to provide a roadmap that will eventually evolve into an e-governance system. If deployed, this proposed system will help continual e-governance data collection. This web based system will collect data from mobile devices that are available at locations of data collection. The second part of this system is to summarize and visualize the data collected. The visualization system is map based so that the policy makers can interact with the system from different views that are relevant for planning and resource allocation. [ABSTRACT FROM AUTHOR]

This paper presents a study about consumers requirements of mobile applications. More specifically, mobile camera related uses. An application was developed because Mobile Phone Camera users demanded a faster photo taking facility, especially while user requires use of mobile camera when they are moving. The paper also presents how and why people adapted the application in their daily lives. The application when launched will start taking photos automatically in the Burst mode (capturing a series of photos one after the other) until the user manually stops it. [ABSTRACT FROM AUTHOR]

In this paper, we propose a novel algorithm that can classify the electroencephalogram (EEG) signals extracted by a single sensor EEG device. The data are captured when subjects were doing three different tasks like - reading/writing texts, solving math problems and watching videos. Our proposed algorithm is capable of extracting the useful information as feature from an EEG signal. The Discrete Wavelet Transform (DWT) is performed to extract frequency band information. To reduce the dimensionality of the higher dimensional feature, we have used Principle Component Analysis (PCA). The K-nearest neighbor classifier is applied finally for EEG signal classification. Our novel approach is capable to operate with 73.12% accuracy. [ABSTRACT FROM AUTHOR]

Diabetic Retinopathy (DR) is a common complication for diabetic patients. It might cause sight loss if not treated early. There are several stages of this disease. Fundus imagery is required to identify the stage and severity of the disease. Due to the lack of proper dataset of the fundus images and proper annotation, it is very difficult to perform research on this topic. To mitigate the problem, we have collected fundus images from different geographic area of Bangladesh and designed annotation software to store information about the patient and the infection and their locations in the images. Sometimes, it is difficult to select all appropriate pixels of the infected region. To resolve the issue, we have introduced a K nearest neighbor (KNN) based technique to accurately select the region of interest (ROI). [ABSTRACT FROM AUTHOR]

Two facial authentication methods based on two different Gabor phase feature representations are proposed in this paper. In the first proposed scheme, similarity score having the highest classification accuracy is used as threshold of the Gabor filter. In the second method, minimum intra-personal similarity score is used as individual subject's threshold for authentication. Both of these methods have shown high classification capability for our dataset. [ABSTRACT FROM AUTHOR]

Signature (Latin - signare) is a handwritten stylized form of identification of its owner. Often handwritten signatures are generally used in secured identity preservation. An ideal signature recognition system handles image noise as well as that of learning unique patterns in an individual's signature. This paper analyzes the performance of artificial neural network (ANN) architectures and Gaussian support vector machine (SVM) kernel for offline signature recognition scheme that is trained on a distinct feature set extracted from signature images. We investigated the impact of using ANN and SVM on specialized feature set and present comparative analysis of the two. Three distinct features - gradient histogram, dot density and slices were used - yielding testing accuracies of 93.1%, 98% and 85.1% respectively. Using ANN and SVM on this set, a maximum accuracy of 96.57% was achieved over a group of 30 individuals, covering an entire data set of 3000 signatures. [ABSTRACT FROM PUBLISHER]

This paper summarizes a research effort for an off-line signature recognition system. Neural network is used to address this problem because the learning and generalization abilities of NNs enable them to cope up with the diversity and the variation of human signatures. Since neural network have proven performance in other pattern recognition tasks such as character recognition therefore, it is equally suitable for the task of signature recognition. In this paper we present a comparative study of signature recognition comprises of three different neural networks i.e., feed-forward-back propagation neural network, competitive and probabilistic neural network. We have proved by experiment and analysis that probabilistic neural network is best suited to deal with signature recognition problem with an average of 100% accuracy. [ABSTRACT FROM PUBLISHER]

In this paper we have proposed a method for biometric authentication using semi-circular segment around retinal fundus optical disc. It is noted that thickness and density of blood vessels are most prominent in close proximity regions around the optical disc since blood vessels enter retina through optical disc. The idea of using these spatial arrangements of blood vessels as feature provides us with uniqueness. Images in RGB and YCbCr color spaces are tested in two separate experiments. The first experiment (Exp-I) resulted in 84.2% accuracy using RGB and in the second experiment (Exp-II), 89.2% accuracy was achieved, in YCbCr color space. The average time taken is 10.25 seconds. [ABSTRACT FROM PUBLISHER]

In this paper, a dual iris based human identification system that increases the accuracy and the performance of a typical human iris recognition system is proposed. The system detects and then isolates and extracts the iris region from eye image. It then sets the radial and angular resolution for the extracted iris region and maps the circular region into rectangular polar coordinates according to the resolutions. This polar image is then convolved with 1D log Gabor filter and phase of the response is quantized to four levels to generate the binary iris template and its corresponding mask. The Hamming distance between two iris templates is computed to find out if the templates are generated from the same iris or not. The conventional method measures this Hamming Distance using iris images from a single eye. However, the proposed method takes images from both eyes simultaneously for comparison process which shows an increased accuracy and performance. Using the proposed method CASIA Iris database V3, false positive rate and false negative rate were found to be 0% and 9.96% respectively while the overall accuracy was 99.92%. [ABSTRACT FROM PUBLISHER]

An effective and efficient resource allocation policy could benefit the cloud environment by saving cost. To support the continuous load increase, the cloud platform needs to create new virtual machines. However, substantial amount of time is required for the creation and the setup of a virtual machine. Therefore, allocating resources in advance based on prediction models could improve the quality of the service of the cloud platform. In this paper we present time delay neural network and regression methods for predicting future workload in the Grid or Cloud platform. We use real world workload traces to test the performance of our algorithms. We also present an overall evaluation of this approach and its potential benefits for enabling efficient auto-scaling of Cloud user resources. [ABSTRACT FROM PUBLISHER]