IJIGSP Vol. 9, No. 1, Jan. 2017
Cover page and Table of Contents: PDF (size: 186KB)
Positron emission tomography images are used to diagnose, staggering, and monitoring several diseases like cancer and Alzheimer, also, this technique is used in clinical research to help to assess the therapeutic and toxic effects of drugs. However, a main drawback of this modality is the poor spatial resolution due to limiting factors such as positron range, instrumentation limits and the allowable doses of radiotracer for administration to patients. These factors also lead to low signal to noise ratios in the images. In this paper, we proposed to increment the resolution of the image and reduce noise by implementing a super resolution scheme, we proposed to use a hybrid regularization consisting of a TV term plus a Tikhonov term to solve the problem of low resolution and heavy noise. By using an anatomical driven scheme to balance between regularization terms we attain a better resolution image with preservation of small structures like lesions and reduced noise without blurring the edges of images. Experimental results and comparisons with other methods of the state-of-the-art show that our proposed scheme produces better preservation of details without adding artifacts when the resolution factor is increased.[...] Read more.
Image segmentation is the process of dividing an electronic digital image into numerous sub-images. Its objective is to categorize image into various regions in such a way that every potential object in image gets individual sector. Instinctive recognition of diabetic retinopathy wounds, like exudates can provide opportunity to identify certain diseases. Lack of accuracy in these techniques can lead to fatal results because of incorrect treatment. So, there is a great need for automation techniques with high accuracy for retinal disease identification. Several automation techniques have been proposed for retinal image analysis which can detect the exudates in fundus images in more promising manner. The related work has found that the issue of noise in fundus images is ignored in the majority of existing literature. Although Gabor filter bank has shown significant results over available techniques, but it is poor in its speed. Also it is not very efficient for multiple kinds of noises at a same time. Therefore to improve the accuracy of exudate extraction further a Hybrid Gabor filter bank with trilateral based filtering technique is proposed. This filtering will use improved trilateral filtering which enables us to detect exudates even in highly corrupted noisy images.[...] Read more.
Internet of Things is an emerging field wherein a lot of classical approaches can be inculcated. One such approach is found in image processing domain. It is real-time object detection and recognition. Object recognition is considered as a complicated process because the object can be of any shape, size or color. Object detection can be performed with effectiveness by using various prevalent techniques such as Scale Invariant Feature Transform (SIFT), a faster version known as Speeded-Up Robust Features (SURF) and the combination of two very efficient algorithms called as Oriented FAST and Rotated BRIEF (ORB) and so on. Although different techniques are dedicated to the different type of objects. In this paper, an effort has been made to combine the object recognition technique with Internet of Things (IoT) concept. The IoT device acting as an input is the camera that captures the image. The object present in the image is detected and recognized. After that, its information is extracted through the internet and displayed on the screen along with the recognized object. The recognition takes place using the pre-existing database. The database consists of the objects that have salient features which would make the task of recognition unambiguous. The bag of features method is considered in order to make recognition effective. The effective use of Internet of Things is carried out by establishing communication between a camera which acts as an input device and visual output devices. This communication takes place over Internet protocol. In the case of object detection, various parameters such as rotation invariance, scale invariance, intensity change, orientation invariance and partial object detection are also considered to make the system robust. Time consideration is carried out to make the system work in real time.[...] Read more.
Many machine learning algorithms work under the assumption that the training and testing data are drawn from the same distribution. However, in practice the assumption might not hold. Transfer subspace learning algorithms aims at utilizing knowledge gained in source domain to learn a task in target domain. The main objective of this work is to apply transfer subspace learning framework on face recognition task at a distance. In this paper we identify face recognition at distance as a transfer learning problem. We show that if the face recognition task is modeled as transfer learning problem, the overall classification rate is increased significantly compared to traditional brute force approach. We also discuss a data set which is unique and meant to advance this research. The novelty of this work lies in modeling face recognition task at distance as a transfer subspace learning problem.[...] Read more.
Objective of this paper is to propose a new Deblurring method for motion blurred textual images. This technique is based on estimating the blur kernel or the Point Spread Function of the motion blur using Blind Deconvolution method. Motion blur is either due to the movement of the camera or the object at the time of image capture. The point spread function of the motion blur is governed by two parameters length of the motion and the angle of the motion. In this approach we have estimated point spread function for the motion blur iteratively for different values of the length and angle of motion. For every estimated PSF we perform the Deconvolution operation with the blurred image to get the non- blurred or the latent image. Latent image obtained is then feed to an Optical character recognition so that the text in that image can be recognized. Then we calculate the Average Word Confidence for the recognized text. Thus for every estimated Point Spread Function and the obtained latent image we get the value of Average Word Confidence. The Point Spread Function with the highest Average Word Confidence value is the optimal Point Spread Function which can be used to deblur the given textual image. In this method we do not have any prior information about the PSF and only single image is used as an input to the system. This method has been tested with the naturally blurred image taken manually and through the internet as well as artificially blurred image for the evaluation of the results. The implementation of the proposed algorithm has been done in MATLAB.[...] Read more.
The present paper aims at performance evaluation of Doubly Truncated Generalized Laplace Mixture Model and Hierarchical clustering (DTGLMM-H) for image analysis concerned to various practical applications like security, surveillance, medical diagnostics and other areas. Among the many algorithms designed and developed for image segmentation the dominance of Gaussian Mixture Model (GMM) has been predominant which has the major drawback of suiting to a particular kind of data. Therefore the present work aims at development of DTGLMM-H algorithm which can be suitable for wide variety of applications and data. Performance evaluation of the developed algorithm has been done through various measures like Probabilistic Rand index (PRI), Global Consistency Error (GCE) and Variation of Information (VOI). During the current work case studies for various different images having pixel intensities has been carried out and the obtained results indicate the superiority of the developed algorithm for improved image segmentation.[...] Read more.
This paper demonstrates mainly on enhancement of extracted feature and proposes a novel approach for feature level fusion for efficient expression recognition. Extracted Gabor filter magnitude feature vector has been fused with upper face part geometrical features and Gabor phase feature vector has been fused with lower face part geometrical features respectively. Both these high dimensional feature dataset have been projected into low dimensional subspace for de-correlating the feature data redundancy by preserving local and global discriminative features of various expression classes of JAFFE, YALE and FD databases. The effectiveness of subspace of fused dataset has been measured with different dimensional parameters of Gabor filter. The experimental results reveal that performance of the subspace approaches for high dimensional proposed feature level fused dataset yields higher accuracy rates compared to state of art approaches.[...] Read more.