IMTA-4 2013 Abstracts


Full Papers
Paper Nr: 1
Title:

A Single Performance Characteristic for the Evaluation of Seeker Tracking Algorithms

Authors:

Leo Doktorski, Eckart Michaelsen and Endre Repasi

Abstract: This paper presents a single numerical performance characteristic for the evaluation of seeker tracking algorithms. It concentrates on ship IR seeker tracking algorithms. Assessing the threat from guided missiles needs a sound evaluation of their performance. The main goal is to introduce a characteristic which is able to assess the threat for ships depending on various scenario pa-rameters. It is shown that for these applications such a single characteristic is sufficient. In order to achieve this seven popular tracking algorithms are used for this. Synthetic IR image sequences are generated to simulate a large set of attack approaches and assemble sufficient statistics on the behavior of the algorithms. The introduced characteristic can also be used for investigations on algorithms themselves, e.g. for sensitivity analyses and parameter optimization of a single algorithm, and for comparison of different algorithms.

Paper Nr: 2
Title:

Curve Recognition for Underwater Wrecks and Handmade Artefacts

Authors:

Davide Moroni, Maria Antonietta Pascali, Marco Reggiannini and Ovidio Salvetti

Abstract: In the framework of the development of autonomous vehicle in order to perform a survey of extreme environments, such as the seabed, the demand for computer vision to support the on-board decision system is increasing. In particular we devote this work to improve the existing underwater curve detection procedures. We propose a method that statistically highlights archaeological artefacts among its environment, weighting properly the persistence of meaningful curves in the video sequence.To this aim we made use of an existing parameterless algorithm ELSD, suitable for digital image processing (Patraucean et al., 2012).

Paper Nr: 3
Title:

Enhanced Resolution Methods for Improving Image Analysis and Pattern Recognition in Scanning Probe Microscopy

Authors:

Mario D'Acunto, Gabriele Pieri, Marco Righi and Ovidio Salvetti

Abstract: Image acquisition systems integrated with laboratory automation produces multi-dimensional datasets. An effective computational approach to objectively analyzing image datasets is pattern recognition (PR), i.e. a machinelearning approach where the machine finds relevant patterns that distinguish groups of objects after being trained on examples (supervised machine learning). In contrast, the other approach to machine learning and artificial intelligence is unsupervised learning, where the intelligent process finds relevant patterns without relying on prior training examples, usually by using a set of pre-defined rules. In this paper we apply a method derived by usual PR techniques for the recognition of artifacts and noise on images recorded with Atomic Force Microscopy (AFM). The advantage of automatic artifacts recognition could be the implementation of machine learning languages for AFM investigations.

Paper Nr: 4
Title:

Conformed Identification of the Fundamental Matrix in the Problem of a Scene Reconstruction, using Stereo Images

Authors:

V. Fursov and Ye. Goshin

Abstract: This paper deals with the problem of the fundamental matrix identification on the basis of corresponding points on stereo images. It is one of the main problems in a scene reconstruction, using stereo images. This problem is commonly solved by the error-adaptive algorithm RANSAC. In this research, this problem is approached in accordance with a conformed identification principle. The method we propose in this paper ensures higher accuracy of the 3D scene reconstruction.

Paper Nr: 5
Title:

Simple Gestalt Algebra

Authors:

Eckart Michaelsen and Vera V. Yashina

Abstract: The laws of Gestalt perception rule how parts are assembled into a perceived aggregate. This contribution defines them in an algebraic setting. Operations are defined for mirror symmetry and repetition in rows respectively. Deviations from the ideal case are handled using positive and differentiable assessment functions achieving maximal value for the ideal case and approaching zero if the parts mutually violate the Gestalt laws. Practically, these definitions and calculations can be used in two ways: 1. Images with Gestalts can be rendered by using random decisions with the assessment functions as densities; 2. given an image (in which Gestalts are supposed) Gestalt-terms are constructed successively, and the ones with high assessment values are accepted as plausible, and thus rec-ognized.

Paper Nr: 6
Title:

Analysis of Large Long-term Remote Sensing Image Sequence for Agricultural Yield Forecasting

Authors:

Alexander Murynin, Konstantin Gorokhovskiy, Valery Bondur and Vladimir Ignatiev

Abstract: Availability of detailed multi-year remote sensing image sequences allows finding a relation between the measured features of vegetation condition history and agricultural yields. The large image sequence over 10 years is used to build and compare 4 yield prediction models. The models are developed trough gradual addition of complexity. The initial model is based on linear regression using vegetation indices. The final model is non-linear and takes into consideration long-term technological advances in agricultural productivity. The accuracy of models has been estimated using cross-validation method. Further ways for model accuracy improvement have been proposed.

Paper Nr: 7
Title:

Mining Coherent Logical Regularities of Type 2 via Positional Preprocessing

Authors:

Alexander Vinogradov and Yury Laptin

Abstract: A new approach to the development of efficient decision rules based on the use of two types of logical regularities is presented. In the approach positional representation of real values of features is used, and as a result, all conversions of types of logical regularities are performed using fast bit operations.

Paper Nr: 8
Title:

Traffic Sign Detection on GPU using Color Shape Regular Expressions

Authors:

Artem Nikonorov, Maksim Petrov and Pavel Yakimov

Abstract: Regular expression matching on GPU is state-of-the-art technique for the processing of a huge amount of the text information, for example, in network activity analyzers and intrusion prevention systems. This paper proposes a fast and robust algorithm for traffic sign detection, which is based on color shape regular expressions matching through the image pixels. We consider a fast massively-parallel GPU implementation of the color shape regular expression, using deterministic finite automaton and special case of non-deterministic finite automaton. The performance of the proposed localization algorithm is compared with the Hough transform, which is commonly used for traffic sign recognition.

Paper Nr: 9
Title:

Video Segmentation by Event Detection: A Novel One-class Classification Approach

Authors:

Mahesh Venkata Krishna, Paul Bodesheim and Joachim Denzler

Abstract: Segmenting videos into meaningful image sequences of some particular activities is an interesting problem in computer vision. In this paper, a novel algorithm is presented to achieve this semantic video segmentation. The goal is to make the system work unsupervised and generic in terms of application scenarios. The segmentation task is accomplished through event detection in a frame-by-frame processing setup. For event detection, we use a one-class classification approach based on Gaussian processes, which has been proved to be successful in object classification. The algorithm is tested on videos from a publicly available change detection database and the results clearly show the suitability of our approach for the task of video segmentation.

Paper Nr: 11
Title:

The Algebraic and Descriptive Approaches and Techniques in Image Analysis

Authors:

I. B. Gurevich, Yu. O. Trusova and V. V. Yashina

Abstract: The main purpose of this review is to explain and discuss the opportunities and limitations of algebraic, linguistic and descriptive approaches in image analysis. During recent years there was accepted that algebraic techniques, in particular different kinds of image algebras, is the most prospective direction of construction of the mathematical theory of image analysis and of development an universal algebraic language for representing image analysis transforms and image models. So, the main goal of the Algebraic Approach is designing of a unified scheme for representation of objects under recognition and its transforms in the form of certain algebraic structures. It makes possible to develop corresponding regular structures ready for analysis by algebraic, geometrical and topological techniques. Development of this line of image analysis and pattern recognition is of crucial importance for automated image mining and application problems solving, in particular for diversification classes and types of solvable problems and for essential increasing of solution efficiency and quality.

Paper Nr: 12
Title:

Fingerprint Identification Problem: Using Delaunay Triangulation Technique for Model Database Indexing

Authors:

Michael Khachay, Alexander Dremin and Anton Leshko

Abstract: A new modification of Delaunay triangulation of finite minutiae subsets extracted from fingerprints is proposed. Stability examining and numerical analysis results are presented as well.

Paper Nr: 13
Title:

Detecting Precise Iris Boundaries by Circular Shortest Path Method

Authors:

Ivan Matveev and Ivan Simonenko

Abstract: Modified circular shortest path detection method is applied for refining pupil and iris boundaries using given approximate pupil and iris circles. Brightness gradient direction is employed to choose image pixels, which may belong to pupil or iris boundary. Using initial approximate circles allows the method to work in a narrow ring, which contains only single circular contour. Under these conditions the method allows to correctly handle almost all images used for iris recognition tasks and appears to be more precise than human expert in marking the pupil border. The method was tested with public domain iris databases, containing more than 80000 images totally.

Paper Nr: 14
Title:

Complexity and Approximability of Hyperplane Covering Problems

Authors:

Michael Khachay

Abstract: The well known N.Megiddo complexity result for Point Cover Problem on the plane is extended onto $d$-dimensional space (for any fixed $d$). It is proved that Min-$d$PC problem is $L$-reducible to Min-$(d+1)$PC problem, therefore for any fixed $d>1$ there is no PTAS for Min-$d$PC problem, unless $P=NP.$

Paper Nr: 15
Title:

On Automated Recognition of Cloud Types Instructions

Authors:

Nina Aprausheva, Irina Gorlach, Aleksandr Zhelnin and Stanislav Sorokin

Abstract: Results of the recognition of multi-spectral satellite data by an automated classification procedure (ACP) are presented. The procedure is based on the approximation of an unknown probability density of a given set of observations by a multi-dimensional Gaussian mixture. For a given number of mixture components, optimal estimates for unknown parameters are found by the Day-Shlezinger algorithm as such solution of simultaneous likelihood equations, that maximizes the likelihood function. Optimal number of classes is determined by the step-by-step testing of two composite statistical hypotheses. The classification of a set of observations is performed by the Bayes rule. To reduce the calculus number, a preliminary analysis of the structure of the investigated set is carried out, which provides rough estimates of the number of classes and their basic characteristics. Results of automatic classification of the main types of clouds and underlying surface are described.

Paper Nr: 16
Title:

Pressure Ulcers Attributes Image Mining to Support Therapeutic Process: A Research Proposal

Authors:

Rinaldo de Souza Neves, Renato Guadagnin, Dirce B. Gulhem and Levy Aniceto Santana

Abstract: The high number of clients suffering from pressure ulcers (PU) overloads hospital services, which are not equipped with the necessary human and technological resources. One needs to increase the productivity of this work, since health professionals, especially nurses use subjective and inaccurate assessments for PU diagnosis. This paper proposes a system for automatic interpretation of tissue and color to support the treatment of patients with PU.

Paper Nr: 17
Title:

Detection and Identification of Neurons in Images of Microscopic Brain Sections

Authors:

Igor Gurevich, Artem Myagkov, Yuriy Sidorov, Yulia Trusova and Vera Yashina

Abstract: This paper presents a new combined mathematical method, which were proposed, implemented, and experimentally tested for extracting information necessary for modeling and, in future, predicting Parkinson’s disease. The method was developed for extraction “neurons” from microscopic images of brain slices of experimental animals. Then it was adapted for different types of initial data, because unfortunately the quality of initial images depends on skills of the specialist who has done an experiment. Now the method allows one to detect and identify as neurons a set of small informative extended objects with well distinguished (by brightness) oval inclusions. The result is a binary image of the contours of detected objects and their inclusions and a list of characteristics calculated for each detected object. The method is based on the joint application of image processing methods, methods of mathematical morphology, methods of segmentation, and the methods of classification of microscopic images. The method was applied to the following areas of brain: the substantia nigra pars compacta and the arcuate nucleus of hypothalamus.