M Tech Dissertations

Permanent URI for this collectionhttp://drsr.daiict.ac.in/handle/123456789/3

Browse

Subject: search.filters.subject.Algorithms

Search Results

Now showing 1 - 10 of 24
  • Thumbnail Image
    ItemOpen Access
    Super resolution of Covid-19 CT-Scan Images
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2022) Patel, Vaidik Gautam; Gohel, Bakul
    Acquisition of high quality CT images is difficult, because it requires exposing patients to high doses of radiation. Super resolution algorithms can help in over coming this problem and obtain higher spatial resolution in CT images. Much deep learning based architecture have been proposed in the literature to overcome this problem. We perform the task of super resolution on a U-Net and study the effects of 2 preprocessing methods which are scaling and zscore. The evaluation strategy for the super resolution of CT images in the literature uses the Peak Signal to Noise Ratio (PSNR) and Structural Similarity (SSIM), however the results are published for the entire image. This is not a good practice for the evaluation of SR, we propose a novel region based similarity measurement practice and a lung specific or region of interest based similarity measurement. We further bifurcate the SSIM metric into it�s 3 component, i.e. luminance, contrast and structure, and study the impact of super resolution on each of these components.
  • Thumbnail Image
    ItemOpen Access
    Material based Mesh Deformation
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2022) Patel, Darshil S.; tatu, Aditya
    Shape deformation is one of the fundamental techniques in geometry processing. Shape deformation algorithms aim to mimic 3D object deformations on digital representations of objects efficiently. Deformation refers to the change in the shape and size of the object. Rigidity performs a significant role in the deformation. The actual 3D object could have parts with different amounts of rigidity. The shape deformation algorithm should be able to model the rigidity of the entire object and spatially varying rigidity. We propose a rigidity controllable mesh deformation method where the user can specify the rigidity of each part of the object. There are several popular deformation models and algorithms to deform objects. However, none of them explicitly capture the rigidity of the object. Additionally, we try to estimate the rigidity of the object given various deformations of the object.
  • Thumbnail Image
    ItemOpen Access
    PolSAR Image Classification using Complex-Valued CNN and Squeeze-Excitation Network
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2022) Makhija, Shradha Vipinkumar; Mandal, Srimanta
    Terrain classification is one of the most crucial tasks when utilisation of polarimetric SAR images comes into the picture. This work explores the efficiency of various supervised deep learning algorithms that make use of Convolutional Neural Networks in land cover classification of PolSAR images. The goal of this work is to classify terrain into different ground covers such as urban, crops, forests, water, etc., from polarimetric SAR (PolSAR) images. State of the art classification approaches relish the advantage of deep learning techniques. However, conventional techniques such as convolutional neural networks (CNN), developed for optical images are not quite suitable for complexvalued PolSAR images. Hence, in this work, complexvalued CNN is employed to deal with complex values of PolSAR images. Further, the CNN focuses mainly on the spatial relationship within local receptive fields. However, the process entangles the channel correlation with spatial information. To address this issue, we use a squeezeexcitation network (SENet) along with complexvalued CNN to exploit the channel interdependencies. Thus, we utilize spatial as well as channel relationships in our work. This, in turn, helps in reducing the speckle noise in the images. Additionally, this work also tests the effectiveness of data augmentation techniques to increase the size of labeled training set of the three datasets used. This is done using various speckle noise suppression techniques. The experimental results on several datasets justify the importance of both spatial information as well as inter- channel correlation in classifying PolSAR images. The results after applying data augmentation techniques specific to the speckled nature of PolSAR images show improved performance of SENet architectures proposed in this work.
  • Thumbnail Image
    ItemOpen Access
    Testing synchronously communicating distributed systems
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2018) Vishwakarma, Nikita; Bhateja, Puneet
    In our work we are supposed to test whether the generated distributed implementation(or distributed network) behaves similar to that of given specification. We are doing this over synchronously communicating FSMs(Finite State Machines). Synchronous distributed implementation is the set of FSMs and specification is a single known FSM. Commonality in the behaviour of given FSM and DFSM(Distributed Finite State Machine) has been checked by conformance. We are also aimed to define a method for testing the global behaviour of the given network in a distributed and concurrent manner. This is achieved by independently performing some operations in each component FSM and later combining these information to get the information of our distributed network. Further this information of distributed network can be used to match its behaviour with specification FSM.
  • Thumbnail Image
    ItemOpen Access
    Total graphs properties of total graphs and dynamic construction of total graphs
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2018) Sheth, Heli; Muthu, Rahul
    This thesis involves studying total graphs which are auxiliary graphs used to transform the total colouring problem of a graph into vertex colouring problem of the total graph. Not all graphs are total graphs but each graph has a unique total graph. Work has been done on recognizing total graphs. So one of the important problems in total graphs is to convert any given non-total graph into total graph by using minimum number of operations on graphs i.e. dynamic construction of total graphs. The main goal of this research was to devise dynamic algorithms which give the largest sub-graph of the given graph which is total. To start with, we have tried to analyse the parameters like clique number, radius, diameter, etc. of total graphs in terms of the original graph and have obtained many results. Also, we discovered another interesting problem in total graphs which is finding the intersection of graph classes with total graphs. This thesis includes results on realizing many special classes of graphs as total graphs. Also, a better algorithmic approach for finding the largest sub-graph of the given input graph has been suggested. However, the suggested algorithm guarantees to give locally optimal solution but does not guarantee a globally optimal solution. The implementation of the previous algorithm and the algorithm that we have suggested has been done and output analysis has been carried out on a range of inputs.
  • Thumbnail Image
    ItemOpen Access
    Designing an optimized linear adaptive filtering algorithm
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2018) Trivedi, Vishwas; Das, Rajib Lochan
    For any system, sparse or dispersive, z2 proportionate algorithm of Proportionate Normalized Least Mean Square (PtNLMS) class, gives the best convergence rate compare to any other weiner filter based iterative algorithms. Though, to implement this algorithm practically, is impossible due to lack of information about the unknown system weight vector wopt, which has been directly used in the formation of the gain proportionality function in z2 proportionate algorithm. For that to formulate, an approximate adaptive model for proportionate gain function is developed using the fuzzy logic. Then using this new proposed approach, formulated the proportionality gain function for some standard adaptive algorithms like NLMS, PNLMS and their convex combinations. Later on, another algorithm of PtNLMS class is developed by formulating the deterministic adaptive gain function, which gives huge improvement in terms of the convergence rate of adaptive filter and provides better stability compare to some adavanced algorithm like IPNLMS.
  • Thumbnail Image
    ItemOpen Access
    Set labeling of graphs
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2018) Kumar, Lokesh; Muthu, Rahul
    Given a universal set and its subsets, intersection graph can be characterized as the graph with one distinct subset of given universal set for each vertex of the graph and any two non-adjacent vertices have no element common in their respective set. This was first studied by Erdos. For Kneser graph and Petersen graph, adjacency is characterized by disjointness. This motivates us to look at disjointness instead of intersection. This report contains results about asymptotic bounds for valid labeling of some special classes of graphs such as harary graphs, split graphs, bipartite graphs, disjoint complete graphs and complete multipartite graphs. Parameters relevant to study of labeling of vertices of the graphs are minimum label size possible (ILN), minimum universe size possible (USN) and their uniform versions such as UILN and UUSN. We have also proposed one framework to label disconnected graphs.
  • Thumbnail Image
    ItemOpen Access
    The Study of Vertex Coloring Algorithms Using Heuristic Approaches
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2018) Lodha, Pratik; Muthu, Rahul
    Graph vertex coloring is one of the most studied NP-complete optimization problem (READ, 1972) [2]. The problem is that; given a graph G, determine the number of colors required to color G, so that no two adjacent vertices share the same color. And the minimum number of colors required to color graph is known as Chromatic Number and is denoted by ?(G). By using existing properties of eccentricity, BFS, DFS (West, 2000) [3] and graph components we have proposed three new heuristic algorithms to obtain approximated chromatic number of a given graph G. And these approaches are as follows: 1. Eccentricity based coloring 2. DFS based coloring, and 3. Maximum degree based coloring.
  • Thumbnail Image
    ItemOpen Access
    The Study of Cycles in 2-connected Graphs Specifically Odd Graphs
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2018) Thakker, Avani; Muthu, Rahul
    Counting the number of cycles in an undirected graph is a classical problemwhich is known to be intractable and so research on this problem typically focuses on approximation algorithms, special cases, heuristics and some variants of the problem. This problem has been extensively studied for its applications in areas of communication systems, artificial intelligence and signal processing. In Complexity theory, this problem lies in the class of #P-complete problem. There may be exponentially many simple cycles in a graph. We observed growth in the number of cycles by adding ears to a 2-connected graph. As analyzed, the growth was exponential. Counting or finding cycles and paths of graphs like complete graphs, presents no interest, in particular since everything is already known analytically. Hence, we studied the cycle structure in Odd Graphs. We analytically obtained cycle lengths that are certainly present in an odd graph without traversing the graph structure. Further, we added minimal number of edges to an odd graph to make the graph pancyclic.
  • Thumbnail Image
    ItemOpen Access
    Testing Labelled Transition System
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2017) Christian, Ruzvelt; Bhateja, Puneet
    "For a successful interaction among computer systems from distinct manufactures, standardize communication protocol are being generated and specified. Implementation of those protocol are required that conforms to those specification. Testing is a process of checking the accuracy of protocol implementation with respect to their specifications. This is called as the protocol conformance testing. In this thesis we will study the formal approach to protocol conformance testing. The final aim of the thesis is to derive approach for the derivation of useful test sets from formal specifications. The test cases that we have derived should not detect any kind of errors. In addition to it, the test cases that we have derived should be meaningful. Starting to this point chapter 1 gives the introduction of the thesis. The idea of testing are being introduced in chapter 1. Also the difference between testing and verification is being introduced. Chapter 2 introduce labelled transition systems in detail. The detail definition and representation have been introduced in this chapter. In chapter 3 relations are proposed using the standard of observation: the behavior of an implementation is true if all observations made of the implementation by an environment, can be explained from the behavior of the specification. In chapter 4, for a particular relation conf, test generation algorithms are developed. These algorithms are used to derive tests from an LTS specification, which should be complete and correct. An implementation relation that we have discussed in chapter 3 and the test derivation algorithm id chapter 4 assume that the tester and the implementation communicate with each other via synchronous communication. Chapter 5 shows that if the implementation and tester interact via a FIFO buffer than it is not applicable. To demonstrate this a queue operator is define to create asynchronous communication among the tester and the implementation. The queue operator consist two queues, one is an input queue and the othe one is an output queue. Any system that interact asynchronously with its environment is called a queue context. Finally, chapter 6 conclude this thesis. By adding the input queue and outputqueue we have converted behaviour of asynchronous system into synchronous system."