Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

Data Mining - Concepts and Techniques.

This paper is a survey of the theory and methods of photogrammetric bundle adjustment, aimed at potential implementors in the computer vision community. Bundle adjustment is the problem of refining a visual reconstruction to produce jointly optimal structure and viewing parameter estimates. Topics covered include: the choice of cost function and robustness; numerical optimization including sparse Newton methods, linearly convergent approximations, updating and recursive methods; gauge (datum) invariance; and quality control. The theory is developed for general robust cost functions rather than restricting attention to traditional nonlinear least squares.

/pdf/bundle-adjustment-a-modern-synthesis-449tfs1h4g.pdf

Bundle Adjustment - A Modern Synthesis

In our recent work [1], [2], we proposed Point Feature Histograms (PFH) as robust multi-dimensional features which describe the local geometry around a point p for 3D point cloud datasets. In this paper, we modify their mathematical expressions and perform a rigorous analysis on their robustness and complexity for the problem of 3D registration for overlapping point cloud views. More concretely, we present several optimizations that reduce their computation times drastically by either caching previously computed values or by revising their theoretical formulations. The latter results in a new type of local features, called Fast Point Feature Histograms (FPFH), which retain most of the discriminative power of the PFH. Moreover, we propose an algorithm for the online computation of FPFH features for realtime applications. To validate our results we demonstrate their efficiency for 3D registration and propose a new sample consensus based method for bringing two datasets into the convergence basin of a local non-linear optimizer: SAC-IA (SAmple Consensus Initial Alignment).

Fast Point Feature Histograms (FPFH) for 3D registration

http://cvssp-data.eps.surrey.ac.uk/Personal/AdrianHilton/Publications_files/survey_techrep.pdf

A survey of advances in vision-based human motion capture and analysis

The Adaptive Least Squares Correlation is a very potent and flexible technique for all kinds of data matching problems. Here its application to image matching is outlined. It allows for simultaneous radiometric corrections and local geometrical image shaping, whereby the system parameters are automatically assessed, corrected, and thus optimized during the least squares iterations. The various tools of least squares estimation can be favourably utilized for the assessment of the correlation quality. Furthermore, the system allows for stabilization and improvement of the correlation procedure through the simultaneous consideration of geometrical constraints, e.g. the collinearity condition. Some exciting new perspectives are emphasized, as for example multiphoto correlation, multitemporal and multisensor correlation, multipoint correlation, and simultaneous correlation/triangulation.

Adaptive least squares correlation: a powerful image matching technique

Particle Tracking Velocimetry (PTV) is a well-known technique for the determination of velocity vectors within an observation volume. However, for a long time it has rarely been applied because of the intensive effort necessary to measure coordinates of a large number of flow marker particles in many images. With today's imaging hardware in combination with the methods of digital image processing and digital photogrammetry, however, new possibilities have arisen for the design of completely automatic PTV systems. A powerful 3D PTV has been developed in a cooperation of the Institute of Geodesy and Photogrammetry with the Institute of Hydromechanics and Water Resources Management at the Swiss Federal Institute of Technology. In this paper hardware components for 3D PTV systems wil be discussed, and a strict mathematical model of photogrammetric 3D coordinate determination, taking into account the different refractive indices in the optical path, will be presented. The system described is capable of determining coordinate sets of some 1000 particles in a flow field at a time resolution of 25 datasets per second and almost arbitrary sequence length completely automatically after an initialization by an operator. The strict mathematical modelling of the measurement geometry, together with a thorough calibration of the system provide for a coordinate accuracy of typically 0.06 mm in X, Y and 0.18 mm in Z (depth coordinate) in a volume of 200 × 160 × 50 mm3.

Particle tracking velocimetry in three-dimensional flows

The automatic co-registration of point clouds, representing 3D surfaces, is a relevant problem in 3D modeling. This multiple registration problem can be defined as a surface matching task. We treat it as least squares matching of overlapping surfaces. The surface may have been digitized/sampled point by point using a laser scanner device, a photogrammetric method or other surface measurement techniques. Our proposed method estimates the transformation parameters of one or more 3D search surfaces with respect to a 3D template surface, using the Generalized Gauss–Markoff model, minimizing the sum of squares of the Euclidean distances between the surfaces. This formulation gives the opportunity of matching arbitrarily oriented 3D surface patches. It fully considers 3D geometry. Besides the mathematical model and execution aspects we address the further extensions of the basic model. We also show how this method can be used for curve matching in 3D space and matching of curves to surfaces. Some practical examples based on the registration of close-range laser scanner and photogrammetric point clouds are presented for the demonstration of the method. This surface matching technique is a generalization of the least squares image matching concept and offers high flexibility for any kind of 3D surface correspondence problem, as well as statistical tools for the analysis of the quality of final matching results.

http://www2.isikun.edu.tr/personel/akca/devrim/2005__LS3D_ISPRSJfinal_vv.PDF

Least squares 3D surface and curve matching

General Topics and Scene Reconstruction- An Overview of DARPA's Research Program in Automatic Population of Geospatial Databases- A Testbed for the Evaluation of Feature Extraction Techniques in a Time Constrained Environment- The Role of Artificial Intelligence in the Reconstruction of Man-Made Objects from Aerial Images- Scene Reconstruction Research - Towards an Automatic System- Semantic Modelling of Man-Made Objects by Production Nets- From Large-Scale DTM Extraction to Feature Extraction- Building Detection and Reconstruction- 3-D Building Reconstruction with ARUBA: A Qualitative and Quantitative Evaluation- A System for Building Detection from Aerial Images- On the Reconstruction of Urban House Roofs from Aerial Images- Image-Based Reconstruction of Informal Settlements- A Model Driven Approach to Extract Buildings from Multi-View Aerial Imagery- Automated Building Extraction from Digital Stereo Imagery- Application of Semi-Automatic Building Acquisition- On the Integration of Object Modeling and Image Modeling in Automated Building Extraction from Aerial Images- TOBAGO - A Topology Builder for the Automated Generation of Building Models- Crestlines Constribution to the Automatic Building Extraction- Recognizing Buildings in Aerial Image- Above-Ground Objects in Urban Scenes from Medium Scale Aerial Imagery- Digital Surface Models for Building Extraction- Extracting Artificial Surface Objects from Airborne Laser Scanner Data- Interpretation of Urban Surface Models using 2D Building Information- Least Squares Matching for Three Dimensional Building Reconstruction- Assessment of the Effects of Resolution on Automated DEM and Building Extraction- Road Extraction- The Role of Grouping for Road Extraction- Artificial Intelligence in 3-D Feature Extraction- Updating Road Maps by Contextual Reasoning- Fast Robust Tracking of Curvy Partially Occluded Roads in Clutter in Aerial Images- Linear Feature Extraction with 3-D LSB-Snakes- Context-Supported Road Extraction- Map/GIS-Based Methods- Three-Dimensional Description of Dense Urban Areas using Maps and Aerial Images- MOSES: A Structural Approach to Aerial Image Understanding- An Approach for the Extraction of Settlement Areas- Extraction of Polygonal Features from Satellite Images for Automatic Registration: The ARCHANGEL Project- Visualisation- A Set of Visualization Data Needs in Urban Environmental Planning & Design for Photogrammetric Data- A Virtual Reality Model of a Major International Airport- Managing Large 3D Urban Database Contents Supporting Phototexture and Levels of Detail- List of Workshop Participants- Author Index

/pdf/automatic-extraction-of-man-made-objects-from-aerial-and-5bbvmwlxid.pdf

Automatic Extraction of Man-Made Objects from Aerial and Space Images (II)

This paper deals with semi-automatic linear feature extraction from digital images for GIS data capture, where the identification task is pe$ormed manually on a single image, while a special automatic digital module performs the high precision feature tracking in two-dimensional (2-0) image space or even three-dimensional (3-0) object space. A human operator identifies the object from an on-screen display of a digital image, selects the particular class this object belongs to, and provides a very few coarsely distributed seed points. subseq;ently, with th;?sk seed as an approximation of the ~osition and sham the linear feature will be extracted automatically by either a dynamic programming approach or by LSB-S~~~~S [Least-Squares E-spline Snakes). With dynamic programming, the optimization problem is set up as a discrete multistage decision process and is solved by a "timedelayed" algorithm. It ensures global optimality, is numerically stable, and allows for hard constraints to be enforced on the solution. In the least-squares approach, we combine three types of observation equations, one radiometric, formulating the matching of a generic object model with image data, and two that express the internal geometric constraints of a curve and the location of operator-given seed points. The solution is obtained by solving a pair of independent normal equations to estimate the parameters of the spline curve. Both techniques can be used in a monoplotting mode, which combines one image with its underlying DTM. The LSB-S~~~~S approach is also implemented in a multi-image mode, which uses multiple images simultaneously and provides for a robust and mathematically sound full 3D approach. These techniques are not restricted to aerial images. They can be applied to satellite and close-range images as well. The issues related to the mathematical modeling of the proposed methods are discussed and experimental results are shown in this paper too.

Armin Gruen

Papers

Adaptive least squares correlation: a powerful image matching technique

Particle tracking velocimetry in three-dimensional flows

Least squares 3D surface and curve matching

Automatic Extraction of Man-Made Objects from Aerial and Space Images (II)

Semi-Automatic Linear Feature Extraction by Dynamic Programming and LSB-Snakes