IJIGSP Vol. 8, No. 3, Mar. 2016
Cover page and Table of Contents: PDF (size: 217KB)
In this study; values obtained through the analysis of blood samples, taken under laboratory conditions, from patients diagnosed with fibromyalgia syndrome and healthy subjects and the sympathetic skin response parameters were used. With the aim of classifying verbal pain scale, which is one of the psychological test scores used for fibromyalgia syndrome diagnosis; relation between the sympathetic skin response effect on other test data and the verbal pain scale were reviewed by using different conditions of available data. Within this framework, three different algorithms were used for classification with high accuracy rates. These algorithms are: Multi-Layer Feed-Forward Neural Networks, Probabilistic Neural Network and Radial Basis Function Neural Network. For Multi-Layer Feed-Forward Neural Networks classification algorithm, classification was done with three different training algorithms, Levenberg-Marquardt back propagation, Resilient back propagation and the Scaled conjugate gradient back propagation and the results were compared elaborately. Based on the results, by using all variables the following accuracy rates were obtained: 68.2% accuracy with Levenberg-Marquardt training algorithm, 77.3% accuracy with the Resilient back propagation training algorithm, and 68.18% accuracy with the Scaled conjugate gradient training algorithm. These success rates show that there is a relationship between verbal pain scale, sympathetic skin response and other test data.[...] Read more.
To select the long-running videos from online archives and other collections, the users would like to browse, or skim through quickly to get a hint on the semantic content of the videos. Video summarization addresses this problem by providing a short video summary of a full-length video. An ideal video summary would include all the important segments of the video and remain short in length. The problem of summarization is extremely challenging and has been a widely pursued subject of recent research. There are many algorithms presented in literature for video summarization and they represent visual information of video in concise form. Dynamic summaries are constructed with collection of key frames or some smaller segments extracted from video and is presented in the form of small video clip. This paper describes an algorithm for constructing the dynamic summary of a video by modeling every 40 consecutive frames of video as a bipartite graph. The method considers every 20 consecutive frames from video as one set and next 20 consecutive frames as second set of bipartite graph nodes with frames of the video representing nodes of the graph and edges connecting nodes denoting the relation between frames and edge weight depicting the mutual information between frames. Then the minimum edge weight maximal matching in every bipartite graph (a set of pair wise non-adjacent edges) is found using Hungarian method. The frames from the matchings which are represented by the nodes connected by the edges with weight below some empirically defined threshold and two neighbor frames are taken as representative frames to construct the summary. The results of the experiments conducted on data set containing sports videos taken from YOUTUBE and videos of TRECVID MED 2011 dataset have demonstrated the satisfactory average values of performance parameters, namely Informativeness value of 94 % and Satisfaction value of 92 %. These values and duration (MSD) of summaries reveal that the summaries constructed are significantly concise and highly informative and provide highly acceptable dynamic summary of the videos.[...] Read more.
The ability of the human visual processing system to accommodate and retain clear understanding or identification of patterns irrespective of their orientations is quite remarkable. Conversely, pattern invariance, a common problem in intelligent recognition systems is not one that can be overemphasized; obviously, one's definition of an intelligent system broadens considering the large variability with which the same patterns can occur. This research investigates and reviews the performance of convolutional networks, and its variant, convolutional auto encoder networks when tasked with recognition problems considering invariances such as translation, rotation, and scale. While, various patterns can be used to validate this query, handwritten Yoruba vowel characters have been used in this research. Databases of images containing patterns with constraints of interest are collected, processed, and used to train and simulate the designed networks. We provide extensive architectural and learning paradigms review of the considered networks, in view of how built-in invariance is learned. Lastly, we provide a comparative analysis of achieved error rates against back propagation neural networks, denoising auto encoder, stacked denoising auto encoder, and deep belief network.[...] Read more.
This Image deblurring aims to eliminate or decrease the degradations that has been occurred while the image has been obtained. In this paper, we proposed a unified framework for restoration process by enhancement and more quantified deblurred images with the help of Genetic Algorithm. The developed method uses an iterative procedure using evolutionary criteria and produce better images with most restored frequency-content. We have compared the proposed methods with Lucy-Richardson Restoration method, method proposed by W. Dong  and Inverse Filter Restoration Method; and demonstrated that the proposed method is more accurate by achieving high quality visualized restored images in terms of various statistical quality measures.[...] Read more.
Modern life style of women has made them more vulnerable to breast cancer and it is considered as the largest cause of mortality among women. This paper presents a novel method to classify mammograms into normal ones, with benign and malignant microcalcifications, and with malignant and benign tumors using fractal features derived from fractal dimension. Here, three fractal dimension estimation methods such as Differential Box Counting (DBC), Triangular Prism Surface Area (TPSA) and Blanket methods are used for computing the six fractal features utilized for the classification. The new fractal feature f6 obtained using TPSA method is found to be the best with 100% classification accuracy. The average value of f6 is found to be 0.1110, 0.2875, 0.4743, 0.5271 and 0.8558, for normal, benign masses, benign and malignant microcalcifications and malignant masses respectively. The classification performance of the different features was analyzed using the Receiver Operating Characteristics (ROC).[...] Read more.
Texture deals with the visual properties of an image. Texture analysis plays a dominant role for image segmentation. In texture segmentation, model based methods are superior to model free methods with respect to segmentation methods. This paper addresses the application of multivariate generalized Gaussian mixture probability model for segmenting the texture of an image integrating with hierarchical clustering. Here the feature vector associated with the texture is derived through DCT coefficients of the image blocks. The model parameters are estimated using EM algorithm. The initialization of model parameters is done through hierarchical clustering algorithm and moment method of estimation. The texture segmentation algorithm is developed using component maximum likelihood under Bayesian frame. The performance of the proposed algorithm is carried through experimentation on five image textures selected randomly from the Brodatz texture database. The texture segmentation performance measures such as GCE, PRI and VOI have revealed that this method outperform over the existing methods of texture segmentation using Gaussian mixture model. This is also supported by computing confusion matrix, accuracy, specificity, sensitivity and F-measure.[...] Read more.
In this paper, we propose a novel algorithm based on directional local difference binary patterns useful for content based image indexing and retrieval. The popular and successful method local binary patterns (LBP) codify a pixel, based on the neighborhood gray values around the pixel. Another flavor of LBP is, center symmetric local binary patterns (CS-LBP), which is the base method for our proposed novel algorithm. The proposed method is based on the directional difference between neighboring pixels. The four directional local difference binary patterns (DLDBP) in 0o, 45o, 90o, and 135o directions are proposed. Then, we apply our method on benchmark image database Corel-1k. The proposed DLDBP (Directional Local Difference Binary Patterns) can also be used to represent a video, using a key frame in the video. We apply the proposed directional local difference binary patterns (DLDBP) key frame based algorithm, on a video database, which consists of ten videos of airplane, ten videos of sailing boat , ten videos of car, and ten videos are of war tank. The performance of proposed DLDBP (Directional Local Difference Binary Patterns) is compared with CS-LBP (Central Symmetric Local Binary Patterns) method. The performance of DLDBP key frame based method is compared with volume local binary patterns (VLBP) method.[...] Read more.
Content-based image retrieval is the process of recovering the images that are based on their primitive features such as texture, color, shape etc. The main challenge in this type of retrieval is the gap between low-level primitive features and high-level semantic concepts. This is known as the semantic gap. This paper proposes an interactive approach for optimizing the semantic gap. The primitive features used are HSV histogram, local binary pattern histogram, and color coherence vector histogram. The mapping between primitive features of the image and its semantic concepts is done by involving the user in the feedback loop. Proposed primitive feature extraction method shows improved image retrieval results (Average precision 73.1%) over existing methods. We have proposed an innovative relevance feedback technique in which the concept of prominent features is introduced. On the application of the relevance feedback, only prominent features which are having maximum similarity are utilized. This method reduces the feature length and increases the efficiency. Our own interactive approach for relevance feedback is not only computationally simple and fast but also shows improvement in the retrieval of semantically meaningful relevant images as we go on increasing the iterations.[...] Read more.