Skip to main content

Über dieses Buch

This book contains extended version of selected works that have been discussed and presented in the fourth International Doctoral Symposium on Applied Computation and Security Systems (ACSS 2017) held in Patna, India during March 17-19, 2017. The symposium was organized by the Departments of Computer Science & Engineering and A. K. Choudhury School of Information Technology, both from University of Calcutta in collaboration with NIT, Patna. The International partners for ACSS 2016 had been Ca Foscari University of Venice, Italy and Bialystok University of Technology, Poland.

This bi-volume book has a total of 21 papers divided in 7 chapters. The chapters reflect the sessions in which the works have been discussed during the symposium. The different chapters in the book include works on biometrics, image processing, pattern recognition, algorithms, cloud computing, wireless sensor networks and security systems.





Hand Biometric Verification with Hand Image-Based CAPTCHA

An approach for hand biometric recognition with the hand image-based CAPTCHA verification is presented in this paper. A new method for CAPTCHA generation is implemented based on the genuine and fake hand images which are embedded in a complex textured color background image. The HandCaptcha is a useful application to differentiate between the human and automated scripts. The first level of security is achieved by the HandCaptcha against the malicious threats and attacks. After solving the HandCaptcha correctly, the identity of a person is authenticated based on the contact-less hand geometric verification approach in the second level. A set of 300 unique HandCaptcha is created randomly and solved by at least 100 persons with the accuracy of 98.34%. Next, the left-hand images of the legitimate users are normalized, and sixteen geometric features are computed from every normalized hand. Experiments are conducted on the 200 subjects of the Bosporus left-hand database. Classification accuracy of 99.5% has been achieved using the kNN classifier, and the equal error rate is 3.93%.
Asish Bera, Debotosh Bhattacharjee, Mita Nasipuri

New Approach to Smartwatch in Human Recognition

Nowadays, personal IoT devices and wearable electronics are taking international markets by storm. Each of these devices is equipped with a variety of different sensors. Throughout the wide range of specific models, one of these sensors is accelerometer. Presented work aims to study the possibility of employing onboard accelerometers of smartwatches to perform owners recognition. We introduce a way of checking the time as a behavioral biometric feature. As a part of this effort, dedicated dataset was created. Then classification algorithms were adjusted and tested. Additionally, comparison was done, between above-mentioned method and more traditional approach, which included computer vision.
Paweł Kobojek, Albert Wolant, Khalid Saeed

Retina Tomography and Optical Coherence Tomography in Eye Diagnostic System

Eye diagnostic, two-step method based on retina color image and Optical Coherence Tomography is presented in this paper. A new robust algorithm, by which various eye diseases can be diagnosed, was implemented as an essential part of the work. The approach comprises two steps. The first one deals with the analysis of retina color image. In this stage, an algorithm was implemented to especially search hard exudates. If the algorithm returns positive, it means at least one hard exudate was found. Moreover, it may return hesitant results in the case of changes that look like hard exudates. During the second step, additional analysis of Optical Coherence Tomography image is done. In this stage, the algorithm is looking for confirmation of hard exudates, which were found during the first step. The authors’ approach gives more confidence in the cases of small exudates or initial stages for eye diseases.
Maciej Szymkowski, Emil Saeed, Khalid Saeed

Typing Signature Classification Model for User Identity Verification

Typing pattern is a behavioral trait of user that is simple, less costly, and workable at any place having only computing device. In this paper, n-graph typing signature is built during user profiling based on keyboard usage pattern. The main aim of this paper is to increase inclusion of number of typing features (both temporal and global) during decision generation and to simplify the procedure of considering missing typing patterns (various monographs, digraphs, etc), which are not enrolled before. A modular classification model collection–storage–analysis (CSA) is designed to identify user. Typing signature becomes adaptive in nature through learning from environment. Module 1 is used for pattern acquisition and processing, and module 2 is used for storage, whereas module 3 is used for analysis. Final decision is generated on the basis of evaluated match score and enrolled global parameters. Proposed CSA model is capable to reduce space and time overhead in terms of dynamic pattern acquisition and storage without using any approximation method. A customized editor HCI is designed for physical key-based devices to build our own data set. Proposed CSA model can classify typing signature of valid and invalid user without incurring high overhead.
Tapalina Bhattasali, Rituparna Chaki, Khalid Saeed, Nabendu Chaki

Image Processing


A Novel Technique for Contrast Enhancement of Chest X-Ray Images Based on Bio-Inspired Meta-Heuristics

Chest radiography is considered as one of the most important radiological tools in pulmonary disease diagnosis. Due to the generation of low contrast images of X-ray machines, the detection of the lesions is a difficult issue and prone to error for a radiologist. Hence, a contrast enhancement algorithm is an obvious choice to enhance the contrast of the image, thus increasing the accuracy of detection of the lesions. This paper not only proposes a new algorithm for contrast enhancement of digital chest X-ray images using particle swarm optimization (PSO), but it also introduces a benchmark dataset of digital chest radiographs to justify the supremacy of our proposed algorithm over that of state-of-the-art contrast enhancement algorithms.
Jhilam Mukherjee, Bishwadeep Sikdar, Amlan Chakrabarti, Madhuchanda Kar, Sayan Das

Ultrasound Medical Image Deblurring and Denoising Method Using Variational Model on CUDA

This paper introduces a new variational model on CUDA platform for the restoration (deblurring and denoising) of ultrasound image degraded by additive Gaussian noise and blur effect. In the deblurring step, we apply an inverse algorithm with the fast transform approach. In the denoising step, a total variational model (TVM) using second-order partial anisotropic diffusion equations is used. A unique and stable solution for the proposed model is presented in terms of the Euler–Lagrange equation. Later, an accurate numerical approximation is constituted by the finite-difference-based discretization technique and the parameter dependence of the proposed model is also described. To achieve better acceleration with satisfactory performance, the proposed algorithm is properly devised on the CUDA GPU and compared with a sequential execution of the multicore CPU system. Experimental results and quantitative analysis show that our algorithm is efficient to restore the ultrasound image compared to the state-of-the-art restoration methods.
Biswajit Biswas, Biplab Kanti Sen, Kashi Nath Dey

Line, Word, and Character Segmentation from Bangla Handwritten Text—A Precursor Toward Bangla HOCR

The basic functionalities of optical character recognition (OCR) are to recognize and extract text to digitally editable text from document images. Apart from this, an OCR has other potentials in document image processing such as in automatic document sorter, writer identification/verification. In current situation, various commercially available OCR systems can be found mostly for Roman script. Development of an unconstrained offline handwritten character recognition system is one of the most challenging tasks for the research community. Things get more complicated when we consider Indic scripts like Bangla which contains more than 280 modified and compound characters along with isolated characters. For recognition of handwritten document, the most convenient way is to segment the text into characters or character parts. So line, word and character level segmentation plays a vital role in the development of such a system. In this paper, a scheme for tri-level segmentation (line, word, and character) is presented. Encouraging segmentation results are achieved on a set of 50 handwritten text documents.
Payel Rakshit, Chayan Halder, Subhankar Ghosh, Kaushik Roy

Heterogeneous Face Matching Using ZigZag Pattern of Local Extremum Logarithm Difference: ZZPLELD

A novel methodology for matching of heterogeneous faces, such as sketch-photo and near-infrared (NIR)-visible (VIS) images is proposed here. For heterogeneous face recognition, more emphasis is given to the edge features, which are invariant in different modality images. Since edges are sensitive to illuminations, we present an illumination-invariant image representation called local extremum logarithm difference (LELD). LELD provides illumination-invariant edge features in coarse level. Therefore, a local zigzag binary pattern LZZBP is presented to capture the local variation of LELD, and we call it a zigzag pattern of local extremum logarithm difference (ZZPLELD). We tested the proposed methodology on different sketch-photo and NIR-VIS benchmark databases. In the case of viewed sketches, the rank-1 recognition accuracy of 96.35% is achieved on CUFSF database. In the case of NIR-VIS matching, the rank-1 accuracy of 99.39% is achieved and which is superior to other state-of-the-art methods. We also tested ZZPLELD on illumination variation Extended Yale B database, and rank-1 recognition accuracy of 94.51% is achieved.
Hiranmoy Roy, Debotosh Bhattacharjee

Pattern Recognition


Automatic Extraction and Identification of Bol from Tabla Signal

In Indian classical music, tabla is the most widely used rhythmic instrument. The instrument has two drums. By striking either of the drums, a bol is produced and it forms the basic component of tala (rhythm). In this work, bols are automatically extracted from tabla signal. Subsequently, features are extracted and used for bol identification. Ideally, a bol follows attack-decay-sustain-release (ADSR) model. A bol has a characteristic rise in the initial attack stage, after which it decays to reach a steady energy level. It sustains that level and, finally, releases the energy. Proposed segmentation methodology exploits this phenomenon to extract the bols. Once the bol segments are extracted, low-level spectral features are computed and used for classification. Multilayer perceptron network is used for bol identification. Experiment is successfully carried out with the signals of recitals by different players and also at different tempo. The result shows that proposed methodology performs quite well on diverse collection. Segmentation and identification of bols can act as the foundation for the applications like transcript generation, tala identification.
Rajib Sarkar, Ankita Singh, Anjishnu Mondal, Sanjoy Kumar Saha

Optimum Circle Formation by Autonomous Robots

This paper considers a constrained version of the circle formation problem for a set of asynchronous, autonomous robots on the Euclidean plane. The circle formation problem asks a set of autonomous, mobile robots, initially having distinct locations, to place themselves, within finite time, at distinct locations on the circumference of a circle (not defined a priori), without colliding with each other. The constrained circle formation problem demands that in addition the maximum distance moved by any robot to solve the problem should be minimized. A basic objective of the optimization constrain is that it implies energy savings of the robots. This paper presents results in two parts. First, it is shown that the constrained circle formation problem is not solvable for oblivious asynchronous robots under ASYNC model even if the robots have rigid movements. Then the problem is studied for robots which have O(1) bits of persistent memory. The initial robot configurations, for which the problem is not solvable in this model, are characterized. For other configurations, a distributed algorithm is presented to solve the problem for asynchronous robots. Only one bit of persistent memory is needed in the proposed algorithm.
Subhash Bhagat, Krishnendu Mukhopadhyaya

Genre Fraction Detection of a Movie Using Text Mining

Movie genre plays a significant role in recommendation system as everyone has a liking for movies of specific genres. Nowadays, a Wikipedia (or wiki) page or plot for each movie is maintained on the Web. In this chapter, we propose to use the Wikipedia movie plot for genre fraction detection using text mining techniques. For our purpose, we use the bag-of-words model as topic modeling where the (frequency of) occurrence of each word is used as a feature for training a classifier. We create the corpus for 20 genres with word frequencies 1, 5, and 15 separately. Wikipedia movie plot of 640 movies is used to evaluate the proposed system. A total of 540 movie plots are used for creating corpuses, and the rest 100 are used as a test set. The system performs best on refined corpus with word frequency 15.
Sunil Saumya, Jitendra Kumar, Jyoti Prakash Singh


Weitere Informationen

Premium Partner