Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
European Journal of Scientific Research ISSN 1450-216X / 1450-202X Vol. 131 No 2 April, 2015, pp.175 - 181 http://www.europeanjournalofscientificresearch.com Signature Identification and Recognition using Elman Neural Network Hind Rostom Mohammed Assistant Professor / Computer Science Department Faculty of Mathematics & Computer Science, Kufa University, Iraq E-mail: [email protected] Elaf Jabbar Abdul Razzaq Al-Taee Assistant Lecturer/ Law Department, Faculty of Law Kufa University, Iraq E-mail: [email protected] Abstract The problem of signature recognition, which is a special case of image analysis, describes how to extracts the features efficiently and verifies the signature perfectly. Human signature can be handling as the image and can be Identified by using Elman neural network. This paper deals with the signature using global features extraction and Elman neural network in which the human signature is captured In image acquisition stage, scanner was used as image acquisition device. In image processing, many techniques are used, through which the signature is extracted from the original image, filtered from noises, and image improvements that prepare the signature to the feature extraction stage. Feature extraction stage which global features is implemented and ending with vector of values for feature extraction method that describes the signature image features. Each feature vector is fed Elman neural network which, that will characterize the signature to whom it belongs. Data base contain 500 signatures for 100 person each person have 5 signatures with different orientation, size, deviation, etc. In an Elman network, the weights from the hidden layer to the context layer are set to one and are fixed because the values of the context neurons have to be copied exactly. The initial output weights of the context neurons are equal to half the output range of the other neurons in the network. The Elman network can be trained with gradient descent back propagation and optimization methods. The accuracy of proposed method was is highly promised result and dependable. Keywords: Signature, Neural Networks, Signature Identification and Recognition, Elman neural network. 1. Introduction Signature authentication technology uses the dynamic analysis of a signature to authenticate a person. The technology is based on measuring speed, pressure and angle used by the person when a signature is produced. This technology uses the individual's handwritten signature as a basis for authentication of entities and data. An electronic drawing tablet and stylus are used to record the direction, speed and coordinates of a handwritten signature [1]. Signature Identification and Recognition using Elman Neural Network 176 Signature recognition is the process of verifying the writer’s identity by checking the signature against samples kept in the database. The result of this process is usually between 0 and 1 which represents a fit ratio (1 for match and 0 for mismatch). Signature recognition is used most often to describe the ability of a computer to translate human writing into text. This may take place in one of two ways either by scanning of written text (off-line method) or by writing directly on to a peripheral input device. The first of these recognition techniques, known as Optical Character Recognition (OCR) is the most successful in the main stream. Most scanning suites offer some form of OCR, allowing user to scan handwritten documents and have them translated into basic text documents. OCR is also used by some archivist as a method of converting massive quantities of handwritten historical documents into searchable, easily-accessible digital forms. As signature is the primary mechanism both for authentication and authorization in legal transactions, the need for efficient auto-mated solutions for signature verification has increased [2]. Our approach is to verify an entered signature with the help an Global properties signature, which is obtained from the set of, collected signatures, we have followed the concept of used Mean square error performance function (MSE) and Mean absolute error performance function (MAE) to find out the amount of divergence in between them using Elman neural network and which is defined as discrimination and signature and identify the person concerned. The rest of the paper is organized as follows: Section 2 gives a brief outline of the 2. Feature extraction and Elman neural network, section 3 describes the process of building results and experiments and shows the experimental results obtained using our method. section 4 gives concluding remarks. 2. Related Work Baltzakis et al. (2001) the proposed system is based on global, grid and texture features. Karouni et al. (2011) present a method for Offline Verification of signatures using a set of simple shape based geometric features. Srinivas et al. (2012) Objective of this work is to recognize the hand written digits represented in black-and-white rectangular pixel displays using Elman neural network (ENN).Bhatia et al.(2013) off-line signature recognition & verification using neural network is proposed, where the signature is captured and presented to the user in an image format. 3. Feature Extraction and Elman Neural Network Feature extraction is a general term for methods of constructing combinations of the variables to get around these problems while still describing the data with sufficient accuracy [3] There are different types of features such as global, grid, texture, and local feature. Global features provide information about specific cases concerning the structure of the signature. Many object recognition systems use global features that describe an entire image. Most shape and texture descriptors fall into this category. Such features are attractive because they produce very compact representations of images, where each image corresponds to a point in a high dimensional feature space. As a result, any standard classifier can be used [4]. There are many Global features such as [3.4]: 1. Area of Image The area of image is calculated by summing all the black pixels in the image. 2. Signature Height This feature is calculated by compute signature height and width. Height equal the difference between the highest and lower black pixels for image. 177 Hind Rostom Mohammed and Elaf Jabbar Abdul Razzaq Al-Taee Width equal the difference between the beginning and end black pixels for horizontal coordinate image. 3. Maximum Horizontal Projection The maximum horizontal projection is calculated by summing all pixels along each row and extracts the maximum projection between them. 4. Maximum Vertical Projection The maximum vertical projection is calculated by summing all pixels along each column and extracts the maximum projection between them. 5. Dimension For more details, we study the dimension of the image which represent the degree of complexity of this image. It is calculated from the following relationship: Dimension =Area /(n*m) where (n, m) = size of image The Elman network was introduced by Elman in 1990. In this network a set of context units are introduced_ which are extra input units whose activation values are fed back from the hidden units. Thus the network is very similar to the Jordan network_ except that 1) The hidden units instead of the output units are fed back 2) The extra input units have no self-connections [5]. Elman neural network is feed forward network with an input layer, a hidden layer, an output layer and a special layer called context layer. The output of each hidden neuron is copied into a specific neuron in the context layer. The value of the context neuron is used as an extra input signal for all the neurons in the hidden layer one time step later. In an Elman network, the weights from the hidden layer to the context layer are set to one and are fixed because the values of the context neurons have to be copied exactly [6]. The key element of this paradigm is the structure of the information processing system. It is composed of a large number of highly interconnected processing elements (neurons) working together to solve specific problems. ANNs, its just like people, learn by example. An ANN is designed for a specific application, such as a data classification, through a learning process. Learning in biological systems involves adjustments to the synaptic connections that exist between the neurons. This is true of ANNs as well [5]. 4. Results and Experiments In this section a detailed experimental Signature Identification and Recognition has been presented. We have used Data base contain 500 signatures for 100 person each person have 5 signatures with different orientation, size, deviation, etc. The flowchart for system show in figure (1) and figure (2) show the sample of signature images while figure (3) show the sample of signature images testing for one person in this paper. The next stage is feature extraction concerns finding for Signature images. To be able to recognize Signature automatically. In feature extraction, we generally seek invariance properties so that the extraction process does not vary according to chosen (or specified) conditions. Global features are sensitive to clutter and occlusion. As a result it is either assumed that an image only contains a single object, or that a good segmentation of the object from the background is available [7].Signatures for this research work are taken from kufa university students. Signature Identification and Recognition using Elman Neural Network Figure 1: Flowchart for System Start Read the image file Converting image file from Color Image to Gray Image Gray Image Enhancement Converting Gray Image to Binary Image Global Feature Extraction Apply the Elman Neural Network Calculate MSE Calculate MAE End Figure 2: The Sample of Gray Scale Images 178 179 Hind Rostom Mohammed and Elaf Jabbar Abdul Razzaq Al-Taee Figure 3: Sample of Signature Images Testing for one Person Figure 4 Showed Global Features for one Person each Person have 5 Signatures Figure 4: Signatures for Person 1 Figure 8 showed an Elman Structure neural network [6]. This network has six input neurons, five hidden neurons and six output neurons. There are also five context neurons. The context neurons receive input from the hidden layers and also pass their output to the hidden layers. The context layers always store the output from the hidden layer and relay this information in the next iteration. This allows them to form a sort of short term memory. Figure 8: Elman Structure Neural Network Signature Identification and Recognition using Elman Neural Network 180 Structural neural network and training: In this research was used Elman neural network and that the results of accurate and efficient in signature verification and distinctiveness. Based network in its work on the values of the Global features of the signature. Where the network receives mono a matrix include 6 values are the results of the Global features of the signature. To illustrate the network layers separately: 1. Input layer: the first layer of Elman network, which includes mono matrix and the number of nodes in this layer (6 nodes) the number of values of the results of the Global features of the signature. 2. Hidden layer: the number of nodes (5 nodes) and the return of my time (context layer) and that gives feedback without weight. 3. Output layer: consists of (6 nodes) according to data target (the results of the Global features of the signature). After the stage of network construction phase begins training the network where they are taking the values of the results of the Global features of the input of the network where the network is trained on error rate (0.001) and the number of cycles of (1000) cycle to reach the desired goal (the results of the Global features of the signature) concept of used Mean square error performance function(MSE) and Mean absolute error performance function(MAE) to find out the amount of divergence in between them using Elman neural network and which is defined as discrimination and signature and identify the person concerned. Figure 9: Mean Square Error Performance Function (MSE) and Mean Absolute Error Performance Function (MAE) for Person 1. Figure 12: Elman Neural Network effects for person (1) contain 5 patterns 181 Hind Rostom Mohammed and Elaf Jabbar Abdul Razzaq Al-Taee Conclusion This paper present comparative analysis of performance and accuracy of elman neural network. Applied the Elman neural network on Data base contain (500) signatures for (100) person each person have (5) signatures. The results obtained from this research work shows the following conclusion:1. We conclude that all five models of the signatures of one person to be the values of MSE and MAE very close. This leads to distinguish this signature to this person. 2. Calculate mse is a network performance function. It measures the network's performance according to the mean of squared errors. 3. Calculate mae is a network performance function. It measures network performance as the mean of absolute errors. 4. Accuracy going to obtain 100% has been achieved In future research, a larger signature database will be collected, including multilingual signatures, to investigate the techniques proposed in this paper. With all the results shown above we can conclude that this new approach performs better. Acknowledgment I would like to thank all those people who made this search possible and an unforgettable experience for me. References [1] [2] [3] [4] [5] [6] [7] Debnath Bhattacharyya and Tai-Hoonkim,Signature Recognition using Artificial Neural Network,Advances in Computational Intelligence, Man-Machine Systems and Cybernetics,2008. Prashanth CR, KB Raja, KR Venugopal, LM Patnaik, “Standard Scores Correlation based Offline signature verification system”, International Conference on advances in computing, control and telecommunication Technologies 2009. Timo Dickscheid, “Coding Images with Local Features”, 27 April 2010. H. Baltzakis, N. Papamarkos, “A new signature verification technique based on a two-stage neural network classifier”, 2001. Elman.J. L, Finding structure in time, Cognitive Science, 14, 1990. J.V.S.Srinivas, P. Premchand Hand Written Digit Recognition Using Elman Neural Network, International Journal of Advanced Research in Computer Science and Software Engineering, Volume 2, Issue 5, May 2012. Mohan Mandaogade, Saurabh and Vishal Mhaske,Handwritten Signature Verification And Recognition Using ANN, MPGI National Multi Conference 2012 (MPGINMC-2012) “Advancement in Electronics & Telecommunication Engineering”, Proceedings published by International Journal of Computer Applications, 7-8 April, 2012.