Jaya Sil

Rice disease identification using pattern recognition techniques

The techniques of machine vision are extensively applied to agricultural science, and it has great perspective especially in the plant protection field, which ultimately leads to crops management. The paper describes a software prototype system for rice disease detection based on the infected images of various rice plants. Images of the infected rice plants are captured by digital camera and processed using image growing, image segmentation techniques to detect infected parts of the plants. Then the infected part of the leaf has been used for the classification purpose using neural network. The methods evolved in this system are both image processing and soft computing technique applied on number of diseased rice plants.

Collusion resilient spread spectrum watermarking in M-band wavelets using GA-fuzzy hybridization

This paper proposes a collusion resilient optimized spread spectrum (SS) image watermarking scheme using genetic algorithms (GA) and multiband (M-band) wavelets. M-band decomposition of the host image offers advantages of better scale-space tiling and good energy compactness. This bandpass-like decomposition makes watermarking robust against frequency selective fading-like gain (intelligent collusion) attack. On the other hand, GA would determine threshold value of the host coefficients (process gain i.e. the length of spreading code) selection for watermark casting along with the respective embedding strengths compatible to the gain of frequency response. First, a single bit watermark embedding algorithm is developed using independent and identically distributed (i.i.d) Gaussian watermark. This is further modified to design a high payload system for binary watermark image using a set of binary spreading code patterns. Watermark decoding performance is improved by multiple stage detection through cancelation of multiple bit interference (MBI) effect. Fuzzy logic is used to classify decision magnitudes in multiple group combined interference cancelation (MGCIC) used in the intermediate stage(s). Simulation results show convergence of GA and validate relative performance gain achieved in this algorithm compared to the existing works.

Dynamic discreduction using Rough Sets

Discretization of continuous attributes is a necessary pre-requisite in deriving association rules and discovery of knowledge from databases. The derived rules are simpler and intuitively more meaningful if only a small number of attributes are used, and each attribute is discretized into a few intervals. The present research paper explores the interrelation between discretization and reduction of attributes. A method has been developed that uses Rough Set Theory and notions of Statistics to merge the two tasks into a single seamless process named dynamic discreduction. The method is tested on benchmark data sets and the results are compared with those obtained by existing state-of-the-art techniques. A real life data on TRIP steel is also analysed using the proposed method.

Cluster Validation Using Splitting and Merging Technique

Analysis of voluminous data generated over the years by business houses, genome projects or elsewhere reveals important findings that advances research activity in respective fields. Grouping meaningful relevant data easily identifies patterns performing similar functions. Clustering is one of the most important techniques used for this purpose. However, obtaining correct number of stable clusters is still an unsolved problem. The proposed method, not sensitive to initialization, generates a set of clusters using the input datasets. The clusters are validated using splitting and merging technique in order to obtain optimal set of clusters. It has been tested on electronic shopping data sets and results are discussed.

Multicarrier spread spectrum watermarking for secure error concealment in fading channel

This paper proposes a novel multicarrier spread spectrum (SS) watermarking scheme for the application of image error concealment using multicarrier-code division multiple access (MC-CDMA) with binary phase shift keying (BPSK) transmission in Rayleigh fading channel. The goal is achieved by embedding important information (image digest) which is extracted from the original image itself, and is used to introduce sufficient redundancy in the transmitted image. Half-toning technique is applied to obtain image digest from its low-resolution version. At the decoder side, data demodulation as well as watermark decoding are done using minimum mean square error combining (MMSEC) strategy. The extracted image digest is used to correct the damaged regions. The integration of SS watermarking with the existing SS modulation not only simplifies the design but also offers significant performance improvement for error concealment in fading channel. Authorized users having the knowledge of code patterns for SS watermarking can only perform the error concealment operation and the method is secured. Experimental results duly support the effectiveness of the proposed scheme.

Color Image Segmentation using Type-2 Fuzzy Sets

Domain knowledge of real life problems are often uncertain, imprecise and inexact, therefore create difficulty in decision making while solving by conventional approaches. Among various methods of handling uncertainties, fuzzy logic has been most intensively studied almost over four decades. Fuzzy logic (FL) explores human reasoning power using linguistic terms, which are modeled as type-1 fuzzy sets and represented by membership functions (MF). However, the MF of type-1 fuzzy set is crisp and cannot tackle all kind of uncertainties . Introducing type-2 fuzzy sets, an extension of type-1 simplifies the problem where the MF is itself fuzzy with three dimension representations. Moreover, a type-2 fuzzy set maps elements of a crisp domain to type-1 fuzzy numbers bounded in the range (0, 1). Research on type-2 fuzzy sets is still in its infancy and very recently has been applied on emerging areas that need development of more efficient and robust systems. The aim of the review paper is to describe type-2 fuzzy systems for managing uncertainties, presenting the frontier research areas where type-2 fuzzy logic has been applied and proposes algorithm on application of type-2 fuzzy sets in color image segmentation.

Clustering by multi objective genetic algorithm

The aim of the paper is to study a real coded multi objective genetic algorithm based K-clustering, where K represents the number of clusters, may be known or unknown. If the value of K is known, it is called K-clustering algorithm. The searching power of Genetic Algorithm (GA) is exploited to get for proper clusters and centers of clusters in the feature space to optimize simultaneously intra-cluster distance (Homogeneity) (H) and inter-cluster distances (Separation) (S). Maximization of 1/H and S are the twin objectives of Multi Objective Genetic Algorithm (MOGA) achieved by measuring H and S using Euclidean distance metric, suitable for continuous features (attributes). We have selected 10 data sets from the UCI machine learning repository containing continuous features only to validate the proposed algorithms. All-important steps of algorithms are shown here. At the end, classification accuracies obtained by best chromosomes are shown.

An efficient classifier design integrating rough set and set oriented database operations

Feature subset selection and dimensionality reduction of data are fundamental and most explored area of research in machine learning and data mining domains. Rough set theory (RST) constitutes a sound basis for data mining, can be used at different phases of knowledge discovery process. In the paper, by integrating the concept of RST and relational algebra operations, a new attribute reduction algorithm has been presented to select the minimum set of attributes, called reducts, required for classification of data. Firstly, the conditional attributes are partitioned into different groups according to their score, calculated using projection (@P) and division (@?) operations of relational algebra. The groups based on their scores are sorted in ascending order while the first group contains maximum information is uniquely used for generating the reducts. The non-reduct attributes are combined with the elements of the next group and the modified group is considered for computing the reducts. The process continues until all groups are exhausted and thus a final set of reducts is obtained. Then applying decision tree algorithm on each reduct, decision rule sets are generated, which are later pruned by removing the extraneous components. Finally, by involving the concept of probability theory and graph theory minimum number of rules is obtained used for building an efficient classifier.

Rough Set Approach to Predict the Strength and Ductility of TRIP Steel

Transformation Induced Plasticity (TRIP) gives birth to new generation steels with high strength and good ductility. Both these properties of steel depend on a number of compositional and processing parameters, but till date there exist certain gaps in the understanding of the complex role of each parameters on the microstructure and thus the properties of the steel. Rough Set Theory is employed to derive decision rules that attempt to explain this complex behavior. Applying efficient heuristics, the number of attributes are reduced to form a minimal reduct, and their values are at the same time discretized into linguistic intervals. The derived rules could clearly indicate on the relative importance of the compositional and processing variables.

Simultaneous feature selection and clustering with mixed features by multi objective genetic algorithm

In this paper, we propose a novel evolutionary clustering algorithm for mixed type data numerical and categorical. It is doing clustering and feature selection simultaneously. Feature subset selection improves quality of clustering. It also improves understandability and scalability. It unfastens attraction on numerical or categorical dataset only. K-prototype KP is a well-known partitional clustering algorithm for mixed type data. However, this type of algorithm is sensitive to initialization and may converge to local optima. It is optimizing a single measure only i.e. minimizations of intra cluster distance. We have considered clustering as a multi objective optimization problem MOOP. Minimization of intra cluster distance and maximization of inter cluster distance are two objectives of optimization. Multi objective genetic algorithm MOGA is a well-known algorithm which can be applicable for MOOP to find out near global optimal solution. So in this paper we have developed a hybridized genetic clustering algorithm by combining the global search ability of MOGA and local search ability of KP. Experiments on real-life benchmark datasets from UCI machine learning repository prove the superiority of the proposed algorithm.