i-manager's Journal on Pattern Recognition (JPR)


Volume 1 Issue 3 September - November 2014

Research Paper

Authentication Based on Handwritten Telugu Character Recognition by Integrating Multiple Contexts

Shoba Bindu C*
*Associate Professor, Computer Science and Engineering Department, Jawaharlal Nehru Technological University Anantapur (JNTUA) College of Engineering, India
Bindu, C. S. (2014). Authentication Based on Handwritten Telugu Character Recognition by Integrating Multiple Contexts. i-manager’s Journal on Pattern Recognition, 1(3), 1-7. https://doi.org/10.26634/jpr.1.3.3213

Abstract

Authentication is the process of confirming the identity of the user using the passwords in order to prove the user is genuine or valid. The graphical patterns can be used with maximum feasibility to the users, as they are easy to remember and hard to guess by the others. This proposed scheme presents an effective approach for the recognition of unconstrained handwritten Telugu texts as passwords for the purpose of granting the access. The proposed scheme uses a special symbol apart from Telugu characters as password, thus increasing the complexity of the password. The proposed scheme is analyzed for usability and security.

Research Paper

Hand Shape Coding: A Robust Approach for Human Identity Verification

Shefali Sharma* , Ashutosh Kumar Singh**, Rajiv Saxena***
Shefali Sharma *  Ashutosh Kumar Singh **  Rajiv Saxena ***
*-** Assistant Professor, Department of Electronics and Communication Engineering, Jaypee University of Engineering and Technology, Guna, India.
*** Director, Jaypee University, Anoopshahr, Uttar Pradesh, India.
Sharma, S., Singh, A. K., and Saxena, R. (2014). Hand Shape Coding: A Robust Approach for Human Identity Verification. i-manager’s Journal on Pattern Recognition, 1(3), 8-17. https://doi.org/10.26634/jpr.1.3.3214

Abstract

The hand based human identity verification and identification is an emerging era for the access control from last decades. Generally, an accurate system is designed by using more than one modality such as shape, geometry and texture information of hand, finger and palm which require more computation. In this paper, the authors have proposed an efficient and robust approach for identity verification using only the shape of the hand. They encoded the shape information in two ways i.e. distance and orientation. Wavelet decomposition is applied to capture the most judicial feature description and to reduce the dimension of both distance and orientation coding. The score level fusion is applied to both the scores obtained using wavelet decomposition of distance and orientation coding for genuine and imposter matches. Finally, false rejection rate and true acceptance rate is computed from the fused genuine and imposter scores. The performance of the introduced mechanism is tested over a hand dataset of 50 subjects. The experimental studies point that good verification performance can be achieved by using only the shape features of the hand with low computational complexity.

Research Paper

Multistage Ink Width Independent Writer Verification Using Chain Code Methods

Sharada Laxman Kore* , Shaila Apte**
* Research Scholar, Bharati Vidyapeeth University College of Engineering, Pune, Maharashtra, India.
** Professor, Department of Electronics and Telecommunication, Rajarshi Shahu College of Engineering, University of Pune, India.
Kore, S. L., and Apte, S. (2014). Multistage Ink Width Independent Writer Verification Using Chain Code Methods. i-manager’s Journal on Pattern Recognition, 1(3), 18-22. https://doi.org/10.26634/jpr.1.3.3215

Abstract

The present work reports the result of writer verification under different ink width conditions using chain code methods. It is observed that the style of a writer and writing instrument used, greatly affects the handwriting. The objective of this work is to improve the verification rate under different ink width conditions. In this work, the writer is verified using multistage approach. The writer is verified based on the writing slant at the first stage, the consistency in the writing style at the second stage, and the writing pressure at the third stage. If the writer is not verified correctly at all three stages, then only it is considered as misclassified. The authors tested the proposed multistage method on their own created dataset of 981 writers including two samples using ball pen and two samples using sketch pen. The experimental result shows that the false acceptance rate is 3.75 % on created dataset of 3924 samples. The proposed system improves accuracy with less computational complexity and verification time.

Research Paper

Gradient Texture Classification Based On Age Prediction of Face Images

Basava Raju* , K. Y. Rama Devi**, P. V. Kumar***
*Research Scholar, Jawarharlal Nehru Technological University, Kakinada, A.P, India
**H.O.D, Department of Computer Science and Engineering, Chaitanya Bharathi Institute of Technology, Gandipet, Hyderabad.
***Professor, Department of Computer Science and Engineering, Osmania University, Hyderabad
Raju, K. B., Devi, Y. R., and Kumar, P. V. (2014). Gradient Texture Classification Based on Age Prediction of Face Images. i-manager’s Journal on Pattern Recognition, 1(3), 23-31. https://doi.org/10.26634/jpr.1.3.3216

Abstract

Skin recognition is used in many applications ranging from algorithms for face detection, predicting age, gender classification, and to objectionable image filtering. These data collections are growing rapidly and can therefore be considered as spatial data streams. For data stream classification, time is a major issue. However, these spatial data sets are too large to be classified effectively in a reasonable amount of time using existing methods. In this work, a novel and computational fast algorithm is proposed for predicting age of humans with PeanoCount Tree (P-Tree). The predicting system was developed and tested based on texture features extracted Local Gradient Patterns (LGP) and Gray Level Cooccurrence Matrix (GLCM) to give better and more predicting accuracy with a range of time period. The P-Tree is a spatial data organization that provides a lossless compressed representation of a spatial data set and facilitates efficient classification and other data mining techniques. Using P-tree structure, fast calculation of measurements, such as information gain, can be achieved. The authors compare P-tree decision tree induction classification and a classical decision tree induction method with respect to the speed at which the classifier can be built (and rebuilt when substantial amounts of new data arrive). Experimental results show that the P-tree method is significantly faster than existing classification methods, making the preferred method for mining on spatial data streams.

Review Paper

Survey on Various Types of License Plate Localization, Segmentation & Recognition Algorithms

C.Anantha Reddy* , Shoba Bindu C**
* PG Student, Department of Computer Science and Engineering, Jawaharlal Nehru Technological University, Anantapuramu, India.
** Associate Professor, Department of Computer Science and Engineering, Jawaharlal Nehru Technological University, Anantapur, India.
Reddy, C. A., and Bindu, C. S. (2014). Survey on Various Types of License Plate Localization, Segmentation & Recognition Algorithms. i-manager’s Journal on Pattern Recognition, 1(3), 32-39. https://doi.org/10.26634/jpr.1.3.3217

Abstract

Automatic License Plate Recognition (ALPR) is a challenging area of review due to its importance to a wide variety of its commercial applications. In any ALPR, there are three phases, which are used for license plate recognition. The initial phase is to capture the car image using sensors like camera and extract the license plate image from the input image. The next phase is to segment the license plate for extracting the characters from the image of the license plate which are based on features like color, shape, etc. The final phase is to detect and recognize the segmented characters of the license plate. This paper reviews different types of approaches and its challenges involved in localization, Segmentation and recognition of license plate numbers. Extensive studies are also made and the recognition accuracies are compared. This paper also suggests the best possible combination within all the proposed techniques to achieve higher accuracies with minimum resources.