Digital Image Processing

J. Jebathangam ¹, S. Purushothaman ², P. Rajeswari ¹

Affiliations
1 Mother Teresa Women’s University, Kodaikanal, IN
2 Institute of Technology, Haramaya University, ET

Abstract

This paper presents a combination of wavelet with echo state neural network in identifying the microcalcification (MC) in a mammogram image. Mammogram image is decomposed using Daubauchi wavelet to 5 levels. Statistical features are extracted from the wavelet coefficients. Training of the ESNN/BPA is done using the features as inputs to the network along with a labeling of presence or non-presence of MC. The classification performance of ESNN is compared with back propagation algorithm.

Keywords

Microcalcification, Wavelet, Neural Network, Echo State Neural Network, Backpropagation Algorithm.

Full Text

     

P. Sujatha ¹, S. Purushothaman ², R. Rajeswari ³

Affiliations
1 School of Computing Sciences, Vels University, Chennai, IN
2 Udaya School of Engineering, Kanyakumari – 629 204, Tamil Nadu, IN
3 Mother Therasa University, Kodaikanal, IN

Abstract

The covert communication based on steganography is a challenging technology for governments. Using this most powerful technique terrorists and spies communicate with each other to exchange their plan which is not detected by law enforcement. In order to avoid the misusage of steganographic technique, the government needs to find out some powerful techniques to detect the existence of the hidden data in the digital media. This leads to the concept of steganalysis that is used in many fields such as digital forensics, medical imaging, and journalism. Apart from all modern sciences and technologies, Artificial Neural Network (ANN) plays a vital role in capturing and representing both linear and non-linear relationships. ANN is an intelligent system which enables machines to solve problems like human by extracting and storing the knowledge. Hence to incorporate intelligent method for steganalysis, this paper implements Artificial Neural Network to overcome the drawbacks of the conventional methods. The most powerful Radial basis function algorithm is proposed in this paper since it is more suitable for non-linear data. This paper concentrates on detecting the hidden information for digital forensics application.

Keywords

Artificial Neural Network, Covert Communication, Radial Basis Function (RBF), Steganography, Steganalysis.

Full Text

     

V. Savithri ¹, S. Purushothaman ²

Affiliations
1 Mother Teresa Women's University, Kodaikanal, IN
2 Udaya School of Engineering, Kanyakumari District-629204, IN

Abstract

The objective of this work is to develop and implement an investigation carried out on carotid artery. The proposed method categorizes the carotid artery subjects into normal and diseased subjects. Ultrasound image videos of the artery are used as the data. The frames of the video are processed to know the properties of the artery. In order to extract properties, image processing techniques have been used on each frame. The features extracted from the frames are consolidating to know the conditions of the artery. These features are accurately analysed to know the status of the artery by using intelligent techniques like artificial neural network and fuzzy logic. In artificial neural network using back propagation algorithm and by fuzzy logic system provides higher classification efficiency with minimum training and testing time. It helps in developing medical decision system for ultrasound artery images. It can also be used as secondary observer in clinical decision making.

Keywords

Artery, Boundary Detection, Intima Media Thickness, Neural Network, Ultrasonic.

Full Text

     

Shoba Rani ¹, S. Purushothaman ²

Affiliations
1 Mother Therasa University, Kodaikanal, IN
2 Sun College of Engineering & Technology, Kanyakumari – 629902, Tamil Nadu, IN

Abstract

The work focuses on segmentation of various textures of a given image. In addition, retrieval of the desired component of the image is done. The segmentation and retrieval is based on the texture features of the image. The goal of the work is to develop a promising technique to segment the textures of the image. Basically,we use a set of GABOR filters segmenting the given image and using artificial neural network (ANN) for extracting the relevant information. 

The theme of the work is based on GABOR filters which are based on the famous Gaussian function. Using Gabor filters, does the extractions of features of the various textures in the given image. By using the obtained features, subsequent labeling is done by (Backpropagation algorithm) ANN. The nature of the work involves taking an image texture as input and getting the partitioned or segmented textures as output. It can be very well justified in doing this work as automatic segmentation, identification and classification and as an important role image processing. The future of the work is unlimited.

Keywords

Texture Image Segmentation, Image Retrieval, Gabor Filters, Artificial Neural Network.

Full Text

     

T. Suganthi ¹, S. Purushothaman ²

Affiliations
1 Mother Teresa Women’s University, Kodaikanal, Tamilnadu, IN
2 Sun College of Engineering & Technology, Kanyakumari – 629902, Tamil Nadu, IN

Abstract

Satellite images are used to understand the land information. But it is difficult to identify the area which is below the clouds. So it is very important task to detect and removal the cloud images. The aim is to give the method for automatic detection and removal of cloud and its shadow contamination from satellite images. After detecting the contamination the method is used to replace the data from different images of the same area to minimize the cloud contamination effect. The Adaptive Brightness Threshold Algorithm (ABT) and Counter Propagation Network (CPN) are used for detection. CPN is able to show the better performance than ABT.

Keywords

Cloud Detection, Threshold, Counter Propagation Network, Satellite Images.

Full Text

     

Shoba Rani ¹, S. Purushothaman ²

Affiliations
1 Department of Computer Science, Mother Therasa University, Kodaikanal, IN
2 Sun College of Engineering and Technology, Nagercoil, IN

Abstract

Texture is a ubiquitous experience and can describe a variety of natural phenomena with repetition, such as sound (background noise in a machine room), motion (animal running), visual appearance (surface color and geometry) and human activities (daily lives). Since reproducing the realism of the physical world is a major goal for computer graphics, textures are important for rendering synthetic images and animations. However, as textures are so diverse it is difficult to describe and reproduce them under a common framework. In this paper, new methods for synthesizing textures are presented. The initial part of the paper is concerned with a basic algorithm for reproducing image textures. The limitations of traditional methods can be overcome by the proposed approach based on multi-resolution (search neighborhoods and tree-structured vector quantization) analysis. The paper concerns with various extensions of the basic algorithm; the extensions concentrate on either reproducing textures of different physical phenomena such as motions, or creating textures in novel ways in addition to mimic existing ones.

Keywords

Texture, Wavelet, Multiresolution Analysis and Image Retrieval.

Full Text