Kannan Balakrishnan

Offline handwritten Malayalam Character Recognition based on chain code histogram

Optical Character Recognition plays an important role in Digital Image Processing and Pattern Recognition. Even though ambient study had been performed on foreign languages like Chinese and Japanese, effort on Indian script is still immature. OCR in Malayalam language is more complex as it is enriched with largest number of characters among all Indian languages. The challenge of recognition of characters is even high in handwritten domain, due to the varying writing style of each individual. In this paper we propose a system for recognition of offline handwritten Malayalam vowels. The proposed method uses Chain code and Image Centroid for the purpose of extracting features and a two layer feed forward network with scaled conjugate gradient for classification.

Speech recognition of Malayalam numbers

Digit speech recognition is important in many applications such as automatic data entry, PIN entry, voice dialing telephone, automated banking system, etc. This paper presents speaker independent speech recognition system for Malayalam digits. The system employs Mel frequency cepstrum coefficient (MFCC) as feature for signal processing and Hidden Markov model (HMM) for recognition. The system is trained with 21 male and female voices in the age group of 20 to 40 years and there was 98.5% word recognition accuracy (94.8% sentence recognition accuracy) on a test set of continuous digit recognition task.

Strongly distance-balanced graphs and graph products

A graph G is strongly distance-balanced if for every edge uv of G and every i>=0 the number of vertices x with d(x,u)=d(x,v)-1=i equals the number of vertices y with d(y,v)=d(y,u)-1=i. It is proved that the strong product of graphs is strongly distance-balanced if and only if both factors are strongly distance-balanced. It is also proved that connected components of the direct product of two bipartite graphs are strongly distance-balanced if and only if both factors are strongly distance-balanced. Additionally, a new characterization of distance-balanced graphs and an algorithm of time complexity O(mn) for their recognition, where m is the number of edges and n the number of vertices of the graph in question, are given.

Machine Learning Approach for Prediction of Learning Disabilities in School-Age Children

his paper highlights the two machine learning approa ches, viz. Rough Sets and Decision Trees (DT), for the prediction of Learning Disabilities (LD) in school-age children, with an emphasis on applications of data mining. Learning disability prediction is a very complicated task. By using these two approaches, we can easily and accurately predict LD in any child and also we can determine the best classification method. In this study, in rough sets the attribute reduction and classification are performed using Johnsons reduction algorithm and Naive Bayes algorithm respectively for rule mining and in construction of decision trees, J48 algorithm is used. From this study, it is concluded that, the performance of decision trees are considerably poorer in several important aspects compared to rough sets. It is found that, for selection of attributes, rough sets is very useful especially in the case of inconsistent data and it also gives the information about the attribute correlation which is very important in the case of learning disability.

Significance Of Classification Techniques In Prediction Of Learning Disabilities

The aim of this study is to show the importance of two classification techniques, viz. decision tree and clustering, in prediction of learning disabilities (LD) of school-age children. LDs affect about 10 percent of all children enrolled in schools. The problems of children with specific learning disabilities have been a cause of concern to parents and teachers for some time. Decision trees and clustering are powerful and popular tools used for classification and prediction in Data mining. Different rules extracted from the decision tree are used for prediction of learning disabilities. Clustering is the assignment of a set of observations into subsets, called clusters, which are useful in finding the different signs and symptoms (attributes) present in the LD affected child. In this paper, J48 algorithm is used for constructing the decision tree and K-means algorithm is used for creating the clusters. By applying these classification techniques, LD in any child can be identified.

Computing median and antimedian sets in median graphs

The median (antimedian) set of a profile π=(u1,…,uk) of vertices of a graph G is the set of vertices x that minimize (maximize) the remoteness ∑id(x,ui). Two algorithms for median graphs G of complexity O(n idim(G)) are designed, where n is the order and idim(G) the isometric dimension of G. The first algorithm computes median sets of profiles and will be in practice often faster than the other algorithm which in addition computes antimedian sets and remoteness functions and works in all partial cubes.

On the remoteness function in median graphs

A profile on a graph G is any nonempty multiset whose elements are vertices from G. The corresponding remoteness function associates to each vertex x@?V(G) the sum of distances from x to the vertices in the profile. Starting from some nice and useful properties of the remoteness function in hypercubes, the remoteness function is studied in arbitrary median graphs with respect to their isometric embeddings in hypercubes. In particular, a relation between the vertices in a median graph G whose remoteness function is maximum (antimedian set of G) with the antimedian set of the host hypercube is found. While for odd profiles the antimedian set is an independent set that lies in the strict boundary of a median graph, there exist median graphs in which special even profiles yield a constant remoteness function. We characterize such median graphs in two ways: as the graphs whose periphery transversal number is 2, and as the graphs with the geodetic number equal to 2. Finally, we present an algorithm that, given a graph G on n vertices and m edges, decides in O(mlogn) time whether G is a median graph with geodetic number 2.

Spectral clustering independent component analysis for tissue classification from brain MRI

Abstract A spectral angle based feature extraction method, Spectral Clustering Independent Component Analysis (SC-ICA), is proposed in this work to improve the brain tissue classification from Magnetic Resonance Images (MRI). SC-ICA provides equal priority to global and local features; thereby it tries to resolve the inefficiency of conventional approaches in abnormal tissue extraction. First, input multispectral MRI is divided into different clusters by a spectral distance based clustering. Then, Independent Component Analysis (ICA) is applied on the clustered data, in conjunction with Support Vector Machines (SVM) for brain tissue analysis. Normal and abnormal datasets, consisting of real and synthetic T1-weighted, T2-weighted and proton density/fluid-attenuated inversion recovery images, were used to evaluate the performance of the new method. Comparative analysis with ICA based SVM and other conventional classifiers established the stability and efficiency of SC-ICA based classification, especially in reproduction of small abnormalities. Clinical abnormal case analysis demonstrated it through the highest Tanimoto Index/accuracy values, 0.75/98.8%, observed against ICA based SVM results, 0.17/96.1%, for reproduced lesions. Experimental results recommend the proposed method as a promising approach in clinical and pathological studies of brain diseases.

Equal opportunity networks, distance-balanced graphs, and Wiener game

Abstract Given a graph G and a set X ⊆ V ( G ) , the relative Wiener index of X in G is defined as W X ( G ) = ∑ { u , v } ∈ X 2 d G ( u , v ) . The graphs G (of even order) in which for every partition V ( G ) = V 1 + V 2 of the vertex set V ( G ) such that | V 1 | = | V 2 | we have W V 1 ( G ) = W V 2 ( G ) are called equal opportunity graphs. In this note we prove that a graph G of even order is an equal opportunity graph if and only if it is a distance-balanced graph. The latter graphs are known by several characteristic properties, for instance, they are precisely the graphs G in which all vertices u ∈ V ( G ) have the same total distance D G ( u ) = ∑ v ∈ V ( G ) d G ( u , v ) . Some related problems are posed along the way, and the so-called Wiener game is introduced.

Multi-Query Content based Image Retrieval System using Local Binary Patterns

Content Based Image Retrieval systems open new research areas in Computer Vision due to the high demand of image searching methods. CBIR is the process of finding relevant image from large collection of images using visual queries. The proposed system uses multiple image queries for finding desired images from database. The different queries are connected using logical AND operation. Local Binary Pattern (LBP) texture descriptors of the query images are extracted and those features are compared with the features of the images in the database for finding the desired images. The proposed system is used for retrieving similar human face expressions. The use of multiple queries reduces the semantic gap between low level visual features and high level user expectation. The experimental result shows that, the use of multiple queries has better retrieval performance over single image queries. General Terms Image Processing, Content Based Image Retrieval