Artificial Visual Attention Using Combinatorial Pyramids

Computer vision systems have to deal with thousands, sometimes millions of pixel values from each frame, and the computational complexity of many problems related to the interpretation of image data is very high. The task becomes especially difficult if a system has to operate in real-time. Within the Combinatorial Pyramid framework, the proposed computational model of attention integrates bottom-up and top-down factors for attention. Neurophysiologic studies have shown that, in humans, these two factors are the main responsible ones to drive attention. Bottom-up factors emanate from the scene and focus attention on regions whose features are sufficiently discriminative with respect to the features of their surroundings. On the other hand, top-down factors are derived from cognitive issues, such as knowledge about the current task. Specifically, the authors only consider in this model the knowledge of a given target to drive attention to specific regions of the image. With respect to previous approaches, their model takes into consideration not only geometrical properties and appearance information, but also internal topological layout. Once the focus of attention has been fixed to a region of the scene, the model evaluates if the focus is correctly located over the desired target. This recognition algorithm considers topological features provided by the pre-attentive stage. Thus, attention and recognition are tied together, sharing the same image descriptors.

An Embedded Vision System for RoboCup

In this chapter, the authors describe their vision system used in the Standard Platform League (SPL), one of the official leagues in RoboCup competition. The characteristics of SPL are very demanding, as all the processing must be done on board, and the changeable environment requires powerful methods for extracting information and robust filters. The purpose is to show a vision system that meets these goals. The chapter describes the architecture of the authors’ system as well as the flowchart of the image process, which is designed in such a manner that allows a rapid and reliable calibration. The authors deal with field features detection by finding intersections between field lines at frame rate, using a fuzzy-Markov localisation technique. Also, the methods implemented to recognise the ball and goals are explained.

Computer Vision for Learning to Interact Socially with Humans

Computer vision is essential to develop a social robotic system capable to interact with humans. It is responsible to extract and represent the information around the robot. Furthermore, a learning mechanism, to select correctly an action to be executed in the environment, pro-active mechanism, to engage in an interaction, and voice mechanism, are indispensable to develop a social robot. All these mechanisms together provide a robot emulate some human behavior, like shared attention. Then, this chapter presents a robotic architecture that is composed with such mechanisms to make possible interactions between a robotic head with a caregiver, through of the shared attention learning with identification of some objects.

Computer Vision Applications of Self-Organizing Neural Networks

This chapter aims to address the ability of self-organizing neural network models to manage video and image processing in real-time. The Growing Neural Gas networks (GNG) with its attributes of growth, flexibility, rapid adaptation, and excellent quality representation of the input space makes it a suitable model for real time applications. A number of applications are presented, including: image compression, hand and medical image contours representation, surveillance systems, hand gesture recognition systems, and 3D data reconstruction.

Learning Robot Vision for Assisted Living

This chapter presents an overview of a typical scenario of Ambient Assisted Living (AAL) in which a robot navigates to a person for conveying information. Indoor robot navigation is a challenging task due to the complexity of real-home environments and the need of online learning abilities to adjust for dynamic conditions. A comparison between systems with different sensor typologies shows that vision-based systems promise to provide good performance and a wide scope of usage at reasonable cost. Moreover, vision-based systems can perform different tasks simultaneously by applying different algorithms to the input data stream thus enhancing the flexibility of the system. The authors introduce the state of the art of several computer vision methods for realizing indoor robotic navigation to a person and human-robot interaction. A case study has been conducted in which a robot, which is part of an AAL system, navigates to a person and interacts with her. The authors evaluate this test case and give an outlook on the potential of learning robot vision in ambient homes.

A Review of Registration Methods on Mobile Robots

The task of registering three dimensional data sets with rigid motions is a fundamental problem in many areas as computer vision, medical images, mobile robotic, arising whenever two or more 3D data sets must be aligned in a common coordinate system. In this chapter, the authors review registration methods. Focusing on mobile robots area, this chapter reviews the main registration methods in the literature. A possible classification could be distance-based and feature-based methods. The distance based methods, from which the classical Iterative Closest Point (ICP) is the most representative, have a lot of variations which obtain better results in situations where noise, time, or accuracy conditions are present. Feature based methods try to reduce the great number or points given by the current sensors using a combination of feature detector and descriptor which can be used to compute the final transformation with a method like RANSAC or Genetic Algorithms.

Real-Time Structure Estimation in Dynamic Scenes Using a Single Camera

In this chapter, an online method is developed for estimating 3D structure (with proper scale) of moving objects seen by a moving camera. In contrast to traditionally developed batch solutions for this problem, a nonlinear unknown input observer strategy is used where the object’s velocity is considered as an unknown input to the perspective dynamical system. The estimator is exponentially stable, and hence, provides robustness against modeling uncertainties and measurement noise from the camera. The developed method provides first causal, observer based structure estimation algorithm for a moving camera viewing a moving object with unknown time-varying object velocities.

A Robust Color Watershed Transformation and Image Segmentation Defined on RGB Spherical Coordinates

The representation of the RGB color space points in spherical coordinates allows to retain the chromatic components of image pixel colors, pulling apart easily the intensity component. This representation allows the definition of a chromatic distance and a hybrid gradient with good properties of perceptual color constancy. In this chapter, the authors present a watershed based image segmentation method using this hybrid gradient. Oversegmentation is solved by applying a region merging strategy based on the chromatic distance defined on the spherical coordinate representation. The chapter shows the robustness and performance of the approach on well known test images and the Berkeley benchmarking image database and on images taken with a NAO robot.

Comparative Analysis of Temporal Segmentation Methods of Video Sequences

In this chapter, a comparative analysis of basic segmentation methods of video sequences and their combinations is carried out. Analysis of different algorithms is based on the efficiency (true positive and false positive rates) and temporal cost to provide regions in the scene. These are two of the most important requirements of the design to provide to the tracking with segmentation in an efficient and timely manner constrained to the application. Specifically, methods using temporal information as Background Subtraction, Temporal Differencing, Optical Flow, and the four combinations of them have been analyzed. Experimentation has been done using image sequences of CAVIAR project database. Efficiency results show that Background Subtraction achieves the best individual result whereas the combination of the three basic methods is the best result in general. However, combinations with Optical Flow should be considered depending of application, because its temporal cost is too high with respect to efficiency provided to the combination.

Uniform Sampling of Rotations for Discrete and Continuous Learning of 2D Shape Models

Different methodologies of uniform sampling over the rotation group, SO(3), for building unbiased 2D shape models from 3D objects are introduced and reviewed in this chapter. State-of-the-art non uniform sampling approaches are discussed, and uniform sampling methods using Euler angles and quaternions are introduced. Moreover, since presented work is oriented to model building applications, it is not limited to general discrete methods to obtain uniform 3D rotations, but also from a continuous point of view in the case of Procrustes Analysis.