Image Processing & Applications

Recovering correct or at least realistic colors of underwater scenes is a very challenging issue for imaging techniques, since illumination conditions in a refractive and turbid medium as the sea are seriously altered. The need to correct colors of underwater images or videos is an important task required in all image-based applications like 3D imaging, navigation, documentation, etc. Many imaging enhancement methods have been proposed in literature for these purposes. The advantage of these methods is that they do not require the knowledge of the medium physical parameters while some image adjustments can be performed manually (as histogram stretching) or automatically by algorithms based on some criteria as suggested from computational color constancy methods. One of the most popular criterion is based on gray-world hypothesis, which assumes that the average of the captured image should be gray. An interesting application of this assumption is performed in the Ruderman opponent color space lαβ, used in a previous work for hue correction of images captured under colored light sources, which allows to separate the luminance component of the scene from its chromatic components. In this work, we present the first proposal for color correction of underwater images by using lαβ color space. In particular, the chromatic components are changed moving their distributions around the white point (white balancing) and histogram cutoff and stretching of the luminance component is performed to improve image contrast. The experimental results demonstrate the effectiveness of this method under gray-world assumption and supposing uniform illumination of the scene. Moreover, due to its low computational cost it is suitable for real-time implementation.

Keywords: Color correction , Computational color constancy , Ruderman space , Underwater imaging

First target of this paper is to describe the design the behaviour of the final prototype before its manufacturing of an automotive semi-active differential based on the use of a and to predict its good performances and possible Magnetorheological Fluid (MRF). The MRF allows to control weaknesses. So the costs and the time to market of the new the locking torque and, consequently, to improve the vehicle handling. Second target is to propose a method grounded on a Close-Range Photogrammetry approach for the CAD modeling phase of the device ideated, alternative to the use of the typical Reverse Engineering (RE) techniques. In fact, although a Reverse Engineering process allows the complete 3D reconstruction of the external surfaces and features of a real object, it could often take a lot of time and, in some cases, could be affected by some approximations or errors. Furthermore, a model “reconstructed” could not be the best solution for multiphysics analyses, where the parametric geometry is needed for the modifications of all its features and dimensions for the optimization process in a very short time. For these reasons, in the case studied, the complete CAD prototype, created step by step, is needed and the photogrammetry can represent an interesting solution to enhance the virtual prototyping phase without repercussions on the quality of the results. Starting from the acquisition of particular “key points”, with an acceptable tolerance, the definition of the references (datum axes, curves, planes, centres of holes) needed for the CAD modeling, according to the Top-Down procedure, was possible. Once obtained the preliminary prototype, the final CAD model was created optimizing its dimensions and choosing the adequate materials. To evaluate the goodness of the procedure adopted, the MRF LSD geometry was reconstructed also by means of the Reverse Engineering techniques applied to the physical prototype of the new device ad hoc created. In particular, laser system acquisition and RE dedicated software were used. In this way the comparison between the photogrammetric and Reverse Engineering procedures (in terms of time spent and quality of the results) was allowed and so conducted. Lastly, the results of the FEM analyses carried out to validate the design process and the methodology ideated and adopted were showed.

Keywords: 3D CAD parametric modeling , Automotive differential , Direct modeling , FEM analysis , Magnetorheological fluid , Photogrammetry , Reverse engineering

The problem of dust resuspension in case of Loss Of Vacuum Accident (LOVA) in a nuclear fusion plant (ITER or DEMO like) is an important issue for the safety of workers and the security of environment. The Quantum Electronics and Plasma Physics Research Group has implemented an optical set-up to track dust during a LOVA reproduction inside the experimental facility STARDUST. The shadowgraph technique, in this work, it is applied to track dark dust (like Tungsten). The shadowgraph technique is based on an expanded collimated beam of light emitted by a laser (or a lamp) transversely to the flow field direction. Inside STARDUST the dust moving in the air flow causes variations of refractive index of light that can be detected by the means of a CCD camera. A spatial modulation of the light-intensity distribution on the camera can be measured. The resulting pattern is a shadow of the refractive index field that prevails in the region of the disturbance. The authors use an incandescent white lamp to illuminate the vacuum vessel of STARDUST facility. The light-area passes through the test section that has to be investigated and the images of the dust shadows are collected with a fast CCD camera. The images are then elaborated with mathematical algorithms to obtain information about the velocity fields of dust during the accidents reproduction. The experimental set-up together with a critical analysis of the first results are presented in this paper.

Keywords: Computer vision , Dust , ITER. , Nuclear fusion plants , Radioactive , Safety , Security , Shadowgraph , Tracking

Seagrass meadows are complex ecosystems representing an important source of biodiversity for coastal marine systems, but are subjected to numerous threats from natural and human-based influences. Due to their susceptibility to changing environmental conditions, seagrasses are habitually used in monitoring programmes as biological indicators to assess the ecological status of coastal environments. In this paper we used a non-destructive photo mosaicing technology to quantify seagrass distribution and abundance, and explore benefits of micro-cartographic analysis. Furthermore, the use of photogrammetric tools enhanced the method, which proved to be efficient due to its use of low-cost instruments and its simplicity of implementation. This paper describes the steps required to use this method in meadows of Posidonia oceanica, including: i) camera calibration procedures, ii) programming of video survey, iii) criteria to perform sampling activities, iv) data processing and micro-georeferenced maps restitution, and v) possible study applications.

Keywords: Mapping , Photo mosaicing , Photogrammetry , Posidonia oceanica , Seagrass , Underwater photography

Motion capture of the human body has being performed for decades with a growing number of technologies, aims and application fields; but only recent optical markerless technologies based on silhouette recognition and depth sensors which have been developed for videogames control interface have brought motion capture to a broad diffusion. Actually, nowadays there are low cost hardware and software suitable for a wide range of applications that may vary from entertainment domain (e.g., videogames, virtual characters in movies) to the biomechanical and biomedical domain (e.g., gait analysis or orthopedic rehabilitation) and to a huge number of industrial sectors. In this quick evolving scenario it is hard to tell which technology is the most suitable for any desired goal. The aim of the paper is to answer to this issue by presenting a benchmark analysis that compares RGB and RGB-D technologies used to track performing people in a variety of conditions. In order to contrast the solutions, several different tasks have been selected, simultaneously captured and post-processed exactly in the same way. The test campaign has been designed to evaluate pros and cons according to the most important feature of a motion capture technology, such as volume of acquisition, accuracy of joint position and tracking of fast movements. Actors were asked to perform a number of tasks, among which free movements of arms, legs and full body, gait, and tasks performed interacting with a machine. The number of sensors around the scene and their disposition have been considered as well. We used Sony PS Eye cameras and Microsoft Kinect sensors as hardware solutions and iPisoft for data elaboration. The gathered results are organized, compared and discussed stressing performances and limitations of any combination and, at last, we proposed the best candidate technology for some key applications.

Keywords: Benchmark , Depth cameras , Digital Human Models , Motion capture , RGB , RGB-D

Pilling is a complex property of textile fabrics, representing, for the final user, a non-desired feature to be controlled and measured by companies working in the textile industry. Traditionally, pilling is assessed by visually comparing fabrics with reference to a set of standard images, thus often resulting in inconsistent quality control. A number of methods using machine vision have been proposed all over the world, with almost all sharing the idea that pilling can be assessed by determining the number of pills or the area occupied by the pills on the fabric surface. In the present work a different approach is proposed: instead of determining the number of pills, a machine vision-based procedure is devised with the aim of extracting a number of parameters characterizing the fabric. These are then used to train an artificial neural network to automatically grade the fabrics in terms of pilling. Tested against a set of differently pilled fabrics, the method shows its effectiveness.

Keywords: Computational vision , Image processing , Machine vision system , Neural networks , Pilling , Textile industry

Commonly used to produce the visual effect of full 3D scene on reduced depth supports, bas relief can be successfully employed to help blind people to access inherently bi-dimensional works of art. Despite a number of methods have been proposed dealing with the issue of recovering 3D or 2.5D surfaces from single images, only a few of them explicitly address the recovery problem from paintings and, more specifically, the needs of visually impaired and blind people. The main aim of the present paper is to provide a systematic method for the semi-automatic generation of 2.5D models from paintings. Consequently, a number of ad hoc procedures are used to solve most of the typical problems arising when dealing with artistic representation of a scene. Feedbacks provided by a panel of end-users demonstrated the effectiveness of the method in providing models reproducing, using a tactile language, works of art otherwise completely inaccessible.

Keywords: 2.5D model , Minimization techniques , Shape From Shading , Tactile model

Design of products characterized by high stylistic content and organic shapes in the form of bas-relief (e.g. fashion accessories, commemorative plaques and coins) is traditionally performed starting from handmade drawings or photographs that are manually reproduced by highly skilled craftsmen such as sculptors and engravers and finally digitized by means of 3D scanning. Several Computer-based procedures have been devised with the aim of speeding up this process, which is considerably time consuming, subjective and costly; these are mainly based on image processing techniques such as embossing, enhancement, histogram equalization or dynamic range, also implemented in CAD-based commercial software. However, these approaches are characterized by several limitations preventing them from providing a "correct" final geometry. In view of that, the present work describes a novel method for the creation of digital bas-reliefs from a single image using a Shape From Shading (SFS) based approach with interactive initialization. Image processing-based techniques and minimization SFS methods are first used in order to retrieve a rough version of the objective surface; successively, this is used as initialization for the final reconstruction algorithm. Tested on a set of case studies, the method proved to be effective in providing satisfactory digital bas-relief from single images. © 2013 © 2013 CAD Solutions, LLC.

Keywords: digital bas-relief , image processing , shape from shading , shape retrieval

The use of co-occurrences of patterns in image analysis has been recently suggested as one of the possible strategies to improve on the bag-of-features model. The intrinsically high number of features of the method, however, is a potential limit to its widespread application. Its extension into rotation invariant versions also requires careful consideration. In this paper we present a general, rotation invariant framework for co-occurrences of patterns and investigate possible solutions to the dimensionality problem. Using local binary patterns as bag-of-features model, we experimentally evaluate the potential advantages that co-occurrences can provide in comparison with bag-of-features. The results show that co-occurrences remarkably improve classification accuracy in some datasets, but in others the gain is negligible, or even negative. We found that this surprising outcome has an interesting explanation in terms of the degree of association between pairs of patterns in an image, and, in particular, that the higher the degree of association, the lower the gain provided by co-occurrences in comparison with bag-of-features.

Grey-level co-occurrence matrices (GLCM) have been on the scene for almost forty years and continue to be widely used today. In this paper we present a method to improve accuracy and robustness against rotation of GLCM features for image classification. In our approach co-occurrences are computed through digital circles as an alternative to the standard four directions. We use discrete Fourier transform normalisation to convert rotation dependent features into rotation invariant ones. We tested our method on four different datasets of natural and synthetic images. Experimental results show that our approach is more accurate and robust against rotation than the standard GLCM features. © 2014 Elsevier B.V. All rights reserved.

Keywords: Co-occurrence matrices , Digital circles , Discrete Fourier transform , Rotation invariance , Texture classification

Texture analysis is an area of intense research activity. Like in other fields, the availability of public data for benchmarking is vital to the development of the discipline. In "Texture databases - A comprehensive survey", Hossain and Serikawa recently provided a precious review of a good number of texture datasets, and put an order into this scattered field. The aim of this appendix is to complement the cited work by providing reference to additional image databases of bio-medical textures, textures of materials and natural textures that have been recently employed in experiments with texture analysis. There is in fact a good number of little-known texture databases which have very interesting features, and for this reason are likely to receive increasing attention in the near future. We are convinced that this extension, along with the original article, will be useful to many researchers and practitioners working in the field of texture analysis. © 2014 Elsevier B.V. All rights reserved.

Keywords: Bio-medical images , Database , Materials , Texture

We present a sequential, two-step procedure based on machine vision for detecting and characterizing impurities in paper. The method is based on a preliminary classification step to differentiate defective paper patches (i.e., with impurities) from non-defective ones (i.e., with no impurities), followed by a thresholding step to separate the impurities from the background. This approach permits to avoid the artifacts which occur when thresholding is applied to paper samples that contain no impurities. We discuss and compare different solutions and methods to implement the procedure and experimentally validate it on a datasets of 11 paper classes. The results show that a marked increase in detection accuracy can be obtained with the two-step procedure in comparison with thresholding alone. © 2013 Elsevier B.V. All rights reserved.

Keywords: Image processing , Machine vision , Paper

Remote sensing technologies have been commonly used to perform greenhouse detection and mapping. In this research, stereo pairs acquired by very high-resolution optical satellites GeoEye-1 (GE1) and WorldView-2 (WV2) have been utilized to carry out the land cover classification of an agricultural area through an object-based image analysis approach, paying special attention to greenhouses extraction. The main novelty of this work lies in the joint use of single-source stereo-photogrammetrically derived heights and multispectral information from both panchromatic and pan-sharpened orthoimages. The main features tested in this research can be grouped into different categories, such as basic spectral information, elevation data (normalized digital surface model; nDSM), band indexes and ratios, texture and shape geometry. Furthermore, spectral information was based on both single orthoimages and multiangle orthoimages. The overall accuracy attained by applying nearest neighbor and support vector machine classifiers to the four multispectral bands of GE1 were very similar to those computed from WV2, for either four or eight multispectral bands. Height data, in the form of nDSM, were the most important feature for greenhouse classification. The best overall accuracy values were close to 90%, and they were not improved by using multiangle orthoimages. © 2014 by the authors.

Keywords: GeoEye-1 , Greenhouses , Multiangle image , Normalized digital surface model , Object-based classification , WorldView-2

In this chapter we describe a unifying framework for local binary patterns and variants which we refer to as histograms of equivalent patterns (HEP). In presenting this concept we discuss some basic issues in texture analysis: the problem of defining what texture is; the problem of classifying the many existing texture descriptors; the concept of bag-of-features and the design choices that one has to deal with when designing a texture descriptor. We show how this relates to local binary patterns and related methods and propose a unifying mathematical formalism to express them within the HEP. Finally, we give a geometrical interpretation of these methods as partitioning operators in a high-dimensional space, showing how this representation can propound possible directions for future research. © 2014 Springer-Verlag Berlin Heidelberg.

In the aerospace sector, tolerance allocation and inspection of many components must provide not only for ease of manufacturing and cost reduction but also for reliability and safety requirements. Computer Aided Tolerancing and Inspection (CAT&I) methods based on Reverse Engineering techniques may enhance production quality assessment with significant reduction of the time-to-market. This paper presents an original reverse engineering methodology that is being developed to achieve the complete automation of the inspection process starting from the design requirements (dimensional and geometrical tolerances) as they generally appear in the component's official drawings. After a brief presentation of the proposed methodology and its possible implementation oriented to non-planar surface recognition, an aeronautical component made by beta-forging of Ti6Al4V Titanium Alloy powders is presented as test case. The discussion of the procedure allows to point out its benefits in terms of (a) more efficient management of the inspection process and data consolidation; (b) more thorough comprehension of the real component; and (c) a better understanding of possible feedbacks to be applied in design or manufacturing. Finally some remarks about the limits of the proposed methodology are shown and possible enhancements, that have been already planned to be applied, will be described. © 2014 IEEE.

Keywords: beta-forging , Computer Aided Tolerancing and Inspection , Reverse Engineering , Segmentation , Ti6Al4V Titanium Alloy powders

In the last few decades Experimental Mechanics, helped by advanced technologies to gather 3-D spatial information in non-transparent media, has evolved into a very general tool. It has become possible to observe the internal volume of engineering materials and in the area of biomechanics living internal tissues. This paper contains a brief review of Continuum Mechanics mathematical models that are available to formulate problems in 3-D including large deformations. The extension of the experimental methods that measure displacements in 2-D to 3-D is presented. Two important cases are considered: (a) use of deterministic signals, (b) use of random signals. In order to separate the complexity of the subject of 3-D analysis from the difficulties that arise from the use of random signals, the connection between mathematical models and their experimental determination is presented utilizing deterministic signals. The extension of the use of random signals to the determination of displacements in 2-D to 3-D is outlined. A new method to extract displacement information from random signals is developed and an example of application is provided. Two methods to extract displacement information in 3-D, the classical method based on displacement projections and discrete image correlation (DIC) based on following gradients of intensities are compared. There are many complex steps involved in data processing aside the basic approach, this circumstance makes difficult a comparison between the two methods, however it is possible to conclude that the results are in fair agreement. © The Society for Experimental Mechanics, Inc. 2014.

Keywords: 3-D continuum dynamics , 3-D continuum kinematics , 3-D displacement analysis discrete image correlation (DIC) , 3-D displacement measurement (deterministic, random signals) , Experimental 3-D displacement measurements opaque media , Large deformations

The perception of haptic textures depends on the mechanical interaction between a surface and a biological sensor. A texture is apprehended by sliding one's fingers over the surface of an object. We describe here an apparatus that makes it possible to record the mechanical fluctuations arising from the friction between a human fingertip and easily interchangeable samples. Using this apparatus, human participants tactually scanned material samples. The analysis of the results indicates that the biomechanical characteristics of individual fingertips clearly affected the mechanical fluctuations. Nevertheless, the signals generated for a single material sample under different conditions showed some invariant features. We propose that this apparatus can be a valuable tool for the analysis of natural haptic surfaces.

Keywords: Apparatus , Biomechanics , Biotribology , Humans , Texture

Several methods are available for making metal foams. They allow to tailor their mechanical, thermal, acoustic, and electrical properties for specific applications by varying the relative density as well as the cell size and morphology. Foams have a very heterogeneous structure so that their properties may show a large scatter. In this paper, an aluminum foam produced by means of foaming of powder compacts and another one prepared via the infiltration process were analyzed and compared. Image analysis has been used as a useful tool to determine size, morphology, and distribution of cells in both foams and to correlate cell morphology with the considered manufacturing process. The results highlighted that cell size and morphology are strictly dependent upon the manufacturing method. This paper shows how some standard 2D morphological indicators may be usefully adopted to characterize foams whose structure derives from the specific manufacturing process. © 2013 ASM International.

Keywords: aluminum foams , compact powder process , infiltration process , morphometric analysis

The space debris represents a danger for operative satellites and human missions. A large amount of debris is located in Low Earth Orbit and in Geostationary Orbit. The worldwide surveillance networks, mainly NORAD and RosKosmos, started monitoring debris since a long a time and testify the continuous growing tendency in the number of objects. Since a few years Aerospace Systems Laboratory of University of Rome started optical space debris observation campaigns by dedicated observatories. The observation activity includes not only planning and taking images of selected portions of the sky, but also analysis of raw picture data to extract the relevant astrometry and/or light-curve information. During the optical campaigns a large amount of images is typically taken and one of the most time consuming activities in optical space objects observation is the data analysis, requiring dedicated and specialized man power. This is the reason why the development of automatic images processing algorithms and procedures to identify the presence of debris, to identify its nature and perform the astrometry computations would highly desirable. This paper deals with the development of automatic procedures and algorithms to detect objects with relative motion with respect to stars in both sidereal tracking mode or terrestrial fixed one. In particular, the developed software is able to recognize the debris inside the picture, solve the star field within the picture and use both these information to achieve the angular measurements of the debris. Relevant efforts have been devoted to software development, such that human interaction is not required. The astrometry computations algorithm works without predefined information about the image. In particular the knowledge of the pointing angles are not required, even if starting from a condition close to the actual one improved the convergence speed, and the picture can be solved starting from a "lost in space" condition. The main algorithm drivers and the tradeoffs in the software implementation are depicted and preliminary results and software performance in actual observation campaigns are discussed in the paper.

The space debris is a challenging problem for the human activity in the space. Observation campaigns are conducted around the globe to detect and track uncontrolled space objects. One of the main problems in optical observation is obtaining useful information about the debris dynamical state by the images collected. For orbit determination, the most relevant information embedded in optical observation is the precise angular position, which can be evaluated by astrometry procedures, comparing the stars inside the image with star catalogs. This is typically a time consuming process, if done by a human operator, which makes this task impractical when dealing with large amounts of data, in the order of thousands images per night, generated by routinely conducted observations. An automated procedure is investigated in this paper that is capable to recognize the debris track inside a picture, calculate the celestial coordinates of the image's center and use these information to compute the debris angular position in the sky. This procedure has been implemented in a software code, that does not require human interaction and works without any supplemental information besides the image itself, detecting space objects and solving for their angular position without a priori information. The algorithm for object detection was developed inside the research team. For the star field computation, the software code astrometry.net was used and released under GPL v2 license. The complete procedure was validated by an extensive testing, using the images obtained in the observation campaign performed in a joint project between the Italian Space Agency (ASI) and the University of Bologna at the Broglio Space center, Kenya. © 2014 IAA. Published by Elsevier Ltd. All rights reserved.

Keywords: Automated image analysis , Object detection , Space debris , Star field