Lalitha Rangarajan

Computational approach towards finding evolutionary distance and gene order using promoter sequences of central metabolic pathway

The comparative analysis of motifs of promoter sequences of the genes encoding enzymes of metabolic pathways such as glycolysis and kreb cycle in different genomes can give insights into the understanding of evolutionary and organizational relationships among both the species as well as enzymes. The comparison of resulting analysis with those of the evolutionary distances drawn considering coding regions of the genes allows one to measure the evolution of complete processes. In the present study we have collected promoter sequences of the glycolysis and kreb cycle genes encoding the respective enzymes from the standard EMBL database and extracted ten Transcription factors (TFs) using the TFsearch tool. This information was put together to develop a database CMPP database both offline and online (http://cmpp.sbbiotech.com). The matrix was developed by calculating the distances based on the presence or absence of motifs (TFs). The phylogenetic tree was obtained by using the NJ method by calculating the distances both within and between the enzymes of glycolysis and kreb cycle individually. The present study could also be extended to pathways such as carbohydrate and lipid metabolic networks.

Letters: Diagonal and secondary diagonal locality preserving projection for object recognition

In this paper, the variants of Two Dimensional Locality Preserving Projection (2DLPP) namely Diagonal Locality Preserving Projection (DiaLPP) and Secondary Diagonal Locality Preserving Projection (SDiaLPP) are proposed as the new dimensionality reduction techniques. The 2DLPP method seeks optimal projection vectors by using the row information of the image and the Alternate 2DLPP method seeks optimal projection vectors by using the column information of the image, whereas the DiaLPP seeks optimal projection vectors by interlacing both the rows and column information of the images. Experimental results on subset of COIL object database show that the proposed methods achieves higher recognition rate than 2DLPP and Diagonal Principal Component Analysis(DiaPCA).

Robust Near Duplicate Image Matching for Digital Image Forensics

Local invariant key point extraction has recently emerged as an attractive approach for detecting near duplicate images. Near duplicate images can be: (i) perceptually identical images (e.g. allowing for change in color balance, change in brightness, compression artifacts, contrast adjustment, rotation, cropping, filtering, scaling etc.), (ii) images of the same 3D scene (from different viewpoints). The requirements for identifying near duplicate images vary according to the application. In this paper we focus on image matching strategy that will assist in the detection of forged (copy-paste forgery) images. So far, no specific image matching strategy exists for this application. The state of the art methodologies tend to generate many false positives. In this paper we have introduced a novel matching strategy for pattern matching of key point distributions. Typical experiments conducted with real world images demonstrate success in near duplicate image retrieval for the application of digital image forensic. Proposed method outperforms some of the existing methods and is computationally efficient.

Alignment Free Frequency Based Distance Measures for Promoter Sequence Comparison

With the massive amount of biological sequence data being generated by current technologies there is an urgent need to come out with faster sequence comparison methods. Most of the existing sequence comparison methods are alignment based which are proven to be very computationally complex when compared to the alignment free methods. In this paper, we have proposed alignment free methods for analysis of promoter sequences. Promoter sequences play a crucial role in gene regulation. After extracting the promoter sequence, matrices of motif frequency with position information (Position Specific Motif Matrix (PSMM)) is constructed, this is further taken for promoter analysis. These designed Frequency Based (FD) algorithms are tested on three different promoter datasets obtained from NCBI and UCSC repositories. The results show high similarity values for promoters with similar functionality and low values otherwise.

Symbolic representation and classification of medical X-ray images

In this paper, we propose a symbolic approach for classification of medical X-ray images. Graph cut segmentation is applied to segment the body part of medical X-ray images. A complete directed graph is constructed using the centroid points in the boundary image of the segmented body part image. The complete directed graph is in turn used to extract features of distance and orientation. Further, the boundaries of segmented images are represented by its skeleton end points. Shape features are then extracted from the represented skeleton end points. To assimilate feature variations, we propose to symbolically represent the extracted features of each class in the form of interval valued features. Based on the proposed symbolic representation, symbolic classifier is then used for classification of medical X-ray images. Experimental results reveal the efficiency of our proposed symbolic classification model.

Face Identification from Manipulated Facial Images Using SIFT

Editing on digital images is ubiquitous. Identification of deliberately modified facial images is a new challenge for face identification system. In this paper, we address the problem of identification of a face or person from heavily altered facial images. In this face identification problem, the input to the system is a manipulated or transformed face image and the system reports back the determined identity from a database of known individuals. Such a system can be useful in mugs hot identification in which mugs hot database contains two views (frontal and profile) of each criminal. We considered only frontal view from the available database for face identification and the query image is a manipulated face generated by face transformation software tool available online. We propose SIFT features for efficient face identification in this scenario. Further comparative analysis has been given with well known eigenface approach. Experiments have been conducted with real case images to evaluate the performance of both methods.

Effective Feature Selection for Classification of Promoter Sequences

Exploring novel computational methods in making sense of biological data has not only been a necessity, but also productive. A part of this trend is the search for more efficient in silico methods/tools for analysis of promoters, which are parts of DNA sequences that are involved in regulation of expression of genes into other functional molecules. Promoter regions vary greatly in their function based on the sequence of nucleotides and the arrangement of protein-binding short-regions called motifs. In fact, the regulatory nature of the promoters seems to be largely driven by the selective presence and/or the arrangement of these motifs. Here, we explore computational classification of promoter sequences based on the pattern of motif distributions, as such classification can pave a new way of functional analysis of promoters and to discover the functionally crucial motifs. We make use of Position Specific Motif Matrix (PSMM) features for exploring the possibility of accurately classifying promoter sequences using some of the popular classification techniques. The classification results on the complete feature set are low, perhaps due to the huge number of features. We propose two ways of reducing features. Our test results show improvement in the classification output after the reduction of features. The results also show that decision trees outperform SVM (Support Vector Machine), KNN (K Nearest Neighbor) and ensemble classifier LibD3C, particularly with reduced features. The proposed feature selection methods outperform some of the popular feature transformation methods such as PCA and SVD. Also, the methods proposed are as accurate as MRMR (feature selection method) but much faster than MRMR. Such methods could be useful to categorize new promoters and explore regulatory mechanisms of gene expressions in complex eukaryotic species.

A Method for Segmentation Radiographic Images with Case Study on Welding Defects

Segmentation is one of the most difficult tasks in image processing, particularly in the case of noisy or low contrast images such as radiographic images of welds. In the present study we have segmented defects in radiographic images of weld. The method applied for detecting and discriminating discontinuities in the radiographic weld images. Two Dimensional Left Median Filter (2D-LMF) has been used for enhancing the images. We compared the performance of this method with Mean Shift. Results exhibited the applied method was more effective than Mean Shift in noisy and low contrasted radiographic images of weld.

Source Camera Identification Based on Sensor Readout Noise

A digitally developed image is a viewable image (TIFF/JPG) produced by a camera’s sensor data (raw image) using computer software tools. Such images might use different colour space, demosaicing algorithms or by different post processing parameter settings which are not the one coded in the source camera. In this regard, the most reliable method of source camera identification is linking the given image with the sensor of camera. In this paper, the authors propose a novel approach for camera identification based on sensor’s readout noise. Readout noise is an important intrinsic characteristic of a digital imaging sensor (CCD or CMOS) and it cannot be removed. This paper quantitatively measures readout noise of the sensor from an image using the mean-standard deviation plot, while in order to evaluate the performance of the proposed approach, the authors tested against the images captured at two different exposure levels. Results show datasets containing 1200 images acquired from six different cameras of three different brands. The success of proposed method is corroborated through experiments.

Efficient Enhancement of Microarray Image Using Histogram Specification

In this paper, we propose image enhancement of microarray images using histogram specification method. The proposed approach consists of system model that discuss about finding the type of noise present in the image and enhancing image by removing the noise present in the image. The proposed method is very efficient as it enhances image by revealing most of the microarray spots which is used for subsequent stages of microarray analysis. Performance analysis is done to measure the efficiency by which image is enhanced. The algorithm uses peak signal to noise ratio and mean squared error to quantify the degree of degradation.