Steganography is the technique of hiding secret data within ordinary data by modifying pixel values which appear normal to a casual observer. Steganography which is similar to cryptography helps in secret communication. The cryptography method focuses on the authenticity and integrity of the messages by hiding the contents of the messages. Sometimes, it is not only just enough to encrypt the message but also essential to hide the existence of the message itself. As this avoids misuse of data, this kind of encryption is less suspicious and does not catch attention. To achieve this, Stacked Autoencoder model is developed which initially compresses and encodes the data effectively and which finally decodes the data back from the compressed encoded representation to a representation that is more similar to the original input. The secret data is encrypted using Elliptic Curve Cryptography algorithm and transformed into an image before it is encoded by the model which combines the cover and the secret image. The cover and the secret image produce a container image which after decoding and decrypting gives the secret data. The proposed work consists of multiple networks which are trained with Flickr Image Dataset and results in Secret Image with a Loss of 4% and Container Image with a Loss of 14%.

In this proposed work, an effort has been made to use multiple image files for steganography encoding alongside with the potential of secret text recovery in the event of any image corruption during the stego image transit. Proposed algorithm is effective on the safety factors of secret image, since the embedded checksum will validate for any unauthorized users or intruders, who plan to corrupt the image in any aspect. If any of the stego image underwent any steganalysis or Man in the Middle attack, then this proposed algorithm can effectively identify the potential corruption. The proposed multi cover image steganography model enables the receiver to send the secret text in more secured way and has the ability to detect the corruption in secret message. This solution is not only detecting the possible cover image corruptions but it also withstands one stego image corruption and has the capability to recover the original secret text even after one stego image corruption during the transmission of the secret message. This proposed work will increase the security of secret text that are being sent using steganography methods.

In this research work, proposing an algorithm, which dynamically selects the best compression algorithms among several compression techniques for steganography encoding. Ranking and selection of best algorithm for each and every steganographic transaction is based on several factors like type of cover image being used for the transmission, length of the secret message, type of the message, compression ratio of the secret message being shared, encoding ratio of secret message over the medium etc., The proposed algorithm dynamically ranks and selects the right compression algorithm to be used for the given secret file to occupy lesser embedding space in stego-image.

Due to the development of digital technology large amount of images are stored and shared in internet and personal database. Lot of research has been done in this area. Traditional image retrieval focus on text and content based. It has some drawback while receiving the required informations. AIA is the new interesting research area. It automatically assigns the keywords to an image. In this work fusion of texture feature and shape features are used. Proposed technique used to learns the relationship between image features and keywords. This proposed work focus automatic image retrieval and annotation system with multi class SVM

With the rapid growth of information on the World Wide Web (WWW), classification of web documents has become important for efficient information retrieval. Relevancy of information retrieved can also be improved by considering semantic relatedness between words which is a basic research area in fields of natural language processing, intelligent retrieval, document clustering and classification, word sense disambiguation etc. The web search engine based semantic relationship from huge web corpus can improve classification of documents. This paper proposes an approach for web document classification that exploits information, including both page count and snippets. To identify the semantic relations between the query words, a lexical pattern extraction algorithm is applied on snippets. A sequential pattern clustering algorithm is used to form clusters of different patterns. The page count based measures are combined with the clustered patterns to define the features extracted from the word-pairs. These features are used to train the Support Vector Machine (SVM), in order to classify the web documents. Experimental results demonstrate 5% and 9% improvement in F1 measure for Reuters 21578 and 20 Newsgroup datasets in the classifier performance.

Enterprise Applications are big business applications. They are complex, distributed, scalable, component-based, and large and mission critical. Enterprise applications in cloud are designed to satisfy hundreds of such Enterprise customer needs, but support the same business needs. Application should be capable of supporting multi-user, multi-developer, multi-machine, multi-component that can manipulate massive data and uses parallel-processing methods for processing it. Most enterprise application has sensitive data that requires compliance to security regulations. The data also needs to me masked, in other words encrypted before moving to cloud [1]. Most of the providers today protect data in two ways. One way is to upload the cloud data and then encrypt and other way is to encrypt and then upload data. In first method the keys are maintained by the cloud provider, e.g. Dropbox, Google Drive, Microsoft Sky Drive. The proposed method in this paper is based on the second method. The Encryption mechanism and keys are maintained by the customer. The approaches in previous works will be suitable for point-of-view online back-up a write once and read many times kind of scenario. In case of Enterprise Applications where lot of transactional data is involved, data transfer rate between application and database in cloud should be really faster to have anytime anywhere seamless experience. The proposed method adds algorithms and logics to the existing HCPOD model for fine grained, high performance cloud data access and storage.

This paper proposes a new non-invasive automatic method to detect and classify Diabetes Mellitus (DM) in the patients. The proposed automatic method incorporates tongue image processing and LS-SVM based classifier. Color, Texture and Geometry features of healthy and DM tongue images are extracted and analysed. The optimized values of the above feature are used to train the LS-SVM in order to detect and classify the DM automatically. The proposed methods are tested with a new database and the performance of the proposed method was tested with confusion matrix as it gives 85.38% accuracy.

Segmentation is an essential step in image systems for the accurate lung disease diagnosis, since it delimits lung structures in Computerised Tomography (CT) images. Indeed, image processing techniques can help computer diagnosis if lung region is accurately obtained. A conventional fuzzy c-means clustering algorithm that has been implemented for segmentation of the Computerised Tomography (CT) lung images still suffers with low convergence rate, getting stuck in the local minima and vulnerable to initialization sensitivity. The proposed system presents an intelligent and dynamic approach called Intelligent Fuzzy C-Means (IFCM) to segment the lung nodules automatically and classify the lung nodules effectively using support vector machine classifier. This approach uses the capability of firefly search to find optimal initial cluster centers for the Fuzzy C-Means (FCM) and thus improve the segmentation accuracy. The features are extracted using fused tamura and haralick features after segmentation. These features are trained using different kernels of support vector machine for automatic detection of lung nodules as benign or malignant. The performance of support vector machine is evaluated by computing different measures from confusion matrix.

The problem of web search time complexity and accuracy has been visited in many research papers, and the authors discussed many approaches to improve the search performance. Still the approaches does not produce any noticeable improvement and struggles with more time complexity as well. To overcome the issues identified, an efficient multi mode conceptual clustering algorithm has been discussed in this paper, which identifies the similar interested user groups by clustering their search context according to different conceptual queries. Identified user groups are shared with the related conceptual queries and their results to reduce the time complexity. The multi mode conceptual clustering, performs grouping of search queries and users according to number of users and their search pattern. The concept of search is identified by using Natural language processing methods and the web logs produced by the default web search engines. The author designed a dedicated web interface to collect the web log about the user search and the same data has been used to cluster the social groups according to number of conceptual queries. The search results has been shared between the users of identified social groups which reduces the search time complexity and improves the efficiency of web search in better manner.

Data Storage outsourcing in cloud computing is a rising trend which prompts a number of interesting security issues. Provable data possession (PDP) is a method for ensuring the integrity of data in storage outsourcing. Remote integrity checking is crucial in cloud storage, here using multi cloud. It can help the clients to check their whole outsourced data by without downloading. This research addresses the construction of efficient PDP which called as RSA Based-PDP (RSA-PDP) mechanism for distributed cloud storage to support data migration and scalability of service. Uploading data are stored in different blocks in multicloud. The generation of tags with the length irrelevant to the size of data blocks. To reduce the memory space by using variable length block verification based on hashing algorithm. Cloud service provider is semi honest server so attacker can easily attack data. Here the data are secured from datamining attacker.

The goal of image segmentation is to cluster pixels into salient image regions. Segmentation could be used for object recognition, occlusion boundary estimation within motion or stereo systems, image compression, image editing, or image database lookup. In this paper, we present a color image segmentation using support vector machine (SVM) pixel classification. Firstly, the pixel level color and texture features of the image are extracted and they are used as input to the SVM classifier. These features are extracted using the homogeneity model and Gabor Filter. With the extracted pixel level features, the SVM Classifier is trained by using FCM (Fuzzy C-Means).The image segmentation takes the advantage of both the pixel level information of the image and also the ability of the SVM Classifier. The Experiments show that the proposed method has a very good segmentation result and a better efficiency, increases the quality of the image segmentation compared with the other segmentation methods proposed in the literature. Keywords—Image Segmentation, Support Vector Machine, Fuzzy C–Means, Pixel Feature, Texture Feature, Homogeneity model, Gabor Filter.

Computed Tomography ( CT) lung nodules segmentation is a challenging task for detecting the nodules from images. Medical experts segment the medical images manually, that is almost not clear which leads to false detection. Conventional Fuzzy c-means is unsupervised method that has been implemented for clustering of the CT images still suffers with as low convergence rate, getting stuck in the local minima and vulnerable to initialization sensitivity. Firefly algorithm is a new population-based optimization method is used for solving many complex problems. The proposed system presents a dynamic and intelligent clustering approach called Firefly Search with Fuzzy C-Means ( FSFCM) to segment the lung nodules automatically. This approach uses the capability of firefly search to find optimal initial cluster centers for the FCM and thus improve the segmentation results. The performance of proposed system is compared with other state-of-art algorithms. Experiment is carried out using real time images to investigate the proposed method.

Cloud computing falls into two general categories. Applications being delivered as service and hardware and data centers that provides those services [1]. Cloud storage evolves from just a storage model to a new service model where data is being managed, maintained, and stored in multiple remote severs for back-up reasons. Cloud platform server clusters are running in network environment and it may contain multiple users' data and the data may be scattered in different virtual data centers. In a multi-user shared cloud computing platform users are only logically isolated, but data of different users may be stored in same physical equipment. These equipments can be rapidly provisioned, implemented, scaled up or down and decommissioned. Current cloud providers do not provide the control or at least the knowledge over the provided resources to their customers. The data in cloud is encrypted during rest, transit and back-up in multi tenant storage. The encryption keys are managed per customer. There are different stages of data life cycle Create, Store, Use, Share, Archive and Destruct. The final stage is overlooked [2], which is the complex stage of data in cloud. Data retention assurance may be easier for the cloud provider to demonstrate while the data destruction is extremely difficult. When the SLA between the customer and the cloud provider ends, today in no way it is assured that the particular customers' data is completely destroyed or destructed from the cloud provider's storage. The proposed method identifies way to track individual customers' data and their encryption keys and provides solution to completely delete the data from the cloud provider's multi-tenant storage architecture. It also ensures deletion of data copies as there are always possibilities of more than one copy of data being maintained for back-up purposes. The data destruction proof shall also be provided to customer making sure that the owner's data is completely removed.

 Abstract — The system is designed to show images which are related to the query image. Extracting color, texture, and shape features from an image plays a vital role in content-based image retrieval (CBIR). Initially RGB image is converted into HSV color space due to its perceptual uniformity. From the HSV image, Color features are extracted using block color histogram, texture features using Haar transform and shape feature using Fuzzy C-means Algorithm. Then, the characteristics of the global and local color histogram, texture features through co-occurrence matrix and Haar wavelet transform and shape are compared and analyzed for CBIR. Finally, the best method of each feature is fused during similarity measure to improve image retrieval effectiveness and accuracy.

Support Vector Machines (SVM) is a machine learning method used for classifying the system.It analyses and identifies the categories using the trained data.It is widely used in medical field for diagnosing the disease.The proposed method consists of four phases.They are lung extraction, segmentation of lung region, feature extraction and finally classification of normal, benign and malignancy in the lung.Threat pixel identification with region growing method is used for segmentation of focal areas in the lung.For feature extraction gray level co-occurrence Matrix (GLCM) is been used.Extracted features are classified using different kernels of Support Vector Machine (SVM).The experimentation is performed with the help of real time computer tomography images.

The number of online documents has grown greatly in recent years due to the increase in popularity of World Wide Web (WWW). The main task of assigning a document of corpus to a set of previously fixed categories is known as document classification. The main issues of document classification involve extraction of discriminating features and then classification of documents based on these features. The accuracy of classification can be improved by considering semantic relation between documents. The proposed work uses annotation retrieval wisdom to improve the accuracy of the document retrieval based on semantic relatedness. Shuffled frog leaping (SFL) algorithm promotes the idea of efficient document classification. Annotation uses singular value decomposition (SVD) which helps to obtain a semantic relationship among documents. This facilitates the accurate retrieval of knowledge between individual and various classes of documents.

The advance of computing technologies leads to growth of images in internet and personal databases. To obtain a required image is a challengeable task. For the last two decades large number of researches focusing on image retrieval with CBIR techniques.CBIR system does not allow the users to query images by using semantic meanings. It will automatically extracted by image processing techniques. Semantic gap exists between image low level features and high level features in this content based image retrieval. To bridge this semantic gap AIA technique has been used. Machine learning techniques used to develop automatic image annotation systems. Retrieval of image can be done by using keywords to the images and image is annotated by semantic keywords. In this paper we focus on survey of image retrieval techniques, including of various key aspect of AIA, both feature extraction and semantic learning methods. Finally we report our findings and provide future research directions in the AIA area in the conclusions.

Due to the exponential growth of information on the Internet and the emergent need to organize them, automated categorization of documents into predefined labels has received an ever-increased attention in the recent years for efficient information retrieval. Relevancy of information retrieved can also be improved by considering semantic relatedness between words which is a basic research area in fields like natural language processing, intelligent retrieval, document clustering and classification and word sense disambiguation. The web search engine based semantic relationship from huge web corpus can improve classification of documents. This paper proposes an approach for web document classification that exploits information, including both page count and snippets and also proposes the use of Artificial Bee Colony (ABC) algorithm as a new tool in the classification task. To identify the semantic relations between the query words, a lexical pattern extraction algorithm is applied on snippets. A sequential pattern clustering algorithm is used to form clusters of different documents. The page count based measures are combined with the clustered documents to define the features extracted from the documents. These features are used to train the ABC algorithm, in order to classify the web documents.