The role of computer vision in prosthetic vision

The cost of vision loss worldwide has been estimated at nearly $3 trillion (http://www.amdalliance.org/cost-of-blindness.html). Non-preventable diseases cause a significant proportion of blindness in developed nations and will become more prevalent as people live longer. Prosthetic vision technologies including retinal implants will play an important therapeutic role. Retinal implants convert an input image stream to visual percepts via stimulation of the retina. This paper highlights some barriers to restoring functional human vision for current generation visual prosthetic devices that computer vision can help overcome. Such computer vision is interactive, aiming to restore function including visuo-motor tasks and recognition.     

Image analysis and computer vision: 1990

This paper presents a bibliography of over 1600 references related to computer vision and image analysis, arranged by subject matter. The topics covered include architectures; computational techniques; feature detection, segmentation, and image analysis; matching, stereo, and time-varying imagery; shape and pattern; color and texture; and three-dimensional scene analysis. A few references are also given on related topics, such as computational geometry, computer graphics, image input/output and coding, image processing, optical processing, visual perception, neural nets, pattern recognition, and artificial intelligence.     

Similarity measures in computer vision

A similarity measure is needed in many Computer Vision problems. Although Euclidean distance has traditionally been used, median distance was recently proposed as an alternative, mostly due to its robustness properties. In this paper, a parametric class of distances is presented which allow to introduce a notion of similarity depending on the problem being considered.     

Learning relational structures: Applications in computer vision

We present and compare two new techniques for learning Relational Structures (RSs) as they occur in 2D pattern and 3D object recognition. These techniques, namely, Evidence-Based Networks (EBS-NNets) and Rulegraphs combine techniques from computer vision with those from machine learning and graph matching. The EBS-NNet has the ability to generalize pattern rules from training instances in terms of bounds on both unary (single part) and binary (part relation) numerical features. It also learns the compatibilities between unary and binary feature states in defining different pattern classes. Rulegraphs check this compatibility between unary and binary rules by combining evidence theory with graph theory. The two systems are tested and compared using a number of different pattern and object recognition problems.     

Mathematical statistics and computer vision

In this discussion paper, I present my views on the role on mathematical statistics for solving computer vision problems.     

Bayesian modeling of uncertainty in low-level vision

The need for error modeling, multisensor fusion, and robust algorithms is becoming increasingly recognized in computer vision. Bayesian modeling is a powerful, practical, and general framework for meeting these requirements. This article develops a Bayesian model for describing and manipulating the dense fields, such as depth maps, associated with low-level computer vision. Our model consists of three components: a prior model, a sensor model, and a posterior model. The prior model captures a priori information about the structure of the field. We construct this model using the smoothness constraints from regularization to define a Markov Random Field. The sensor model describes the behavior and noise characteristics of our measurement system. We develop a number of sensor models for both sparse and dense measurements. The posterior model combines the information from the prior and sensor models using Bayes' rule. We show how to compute optimal estimates from the posterior model and also how to compute the uncertainty (variance) in these estimates. To demonstrate the utility of our Bayesian framework, we present three examples of its application to real vision problems. The first application is the on-line extraction of depth from motion. Using a two-dimensional generalization of the Kalman filter, we develop an incremental algorithm that provides a dense on-line estimate of depth whose accuracy improves over time. In the second application, we use a Bayesian model to determine observer motion from sparse depth (range) measurements. In the third application, we use the Bayesian interpretation of regularization to choose the optimal smoothing parameter for interpolation. The uncertainty modeling techniques that we develop, and the utility of these techniques in various applications, support our claim that Bayesian modeling is a powerful and practical framework for low-level vision.     

Color-defective vision and computer graphics displays

A color space defined by the fundamental spectral sensitivity functions of the human visual system is used to assist in the design of computer graphics displays for color-deficient users. The functions are derived in terms of the CIE standard observer color-matching functions. The Farnsworth-Munsell 100-hue test, a widely used color vision test administered using physical color samples, is then implemented on a digitally controlled color television monitor. The flexibility of this computer graphics medium is then used to extend the Farnsworth-Munsell test in a way that improves the specificity of the diagnoses rendered by the test. The issue of how the world appears to color-deficient observers is addressed, and a full-color image is modified to represent a color-defective view of the scene. Specific guidelines are offered for the design of computer graphics displays that will accommodate almost all color-deficient users     

On quantization errors in computer vision

The author considers the error resulting in the computation of multivariable functions h(X₁, X, . . ., X_n), where all the X_is are only available in the quantized form. In image processing and computer vision problems, the variables are typically a mixture of the spatial coordinates and the intensity levels of objects in an image. A method is introduced using a first-order Taylor series expansion together with a periodic extension of the resulting expression and its Fourier series representation so that the moments and the probability distribution function of the error can be computed in closed form. This method only requires that the joint probability density function of X_i s be known and makes no assumption on the behavior on the quantization errors of the variables. Examples are also given where these results are applied     

Estimation of measurement uncertainty in stereo vision system

Stereo vision systems are becoming increasingly popular and widespread. These systems are widely used in many applications, such as navigation of autonomous mobile robots, 3D measurements, object tracking, the movie industry, augmented reality or people tracking and identification systems. Surprisingly, in the literature, little attention is paid to practical verification procedures in which proper operation of the calibrated stereo vision system can be demonstrated. Therefore, in this paper, a novel approach is proposed that allows accurate estimation of the measurement uncertainty of the (x,y,z) coordinates reconstructed by the stereo vision system. The proposed method does not require any additional equipment beyond the standard calibration board and a general-purpose laser distance meter. The authors introduce a simulation model and mathematical formulas that can be employed to determine the accuracy of the stereo vision system precisely along each of the X, Y and Z axes. The measurement uncertainties obtained are statistically reliable because they are calculated with the use of a large amount of data. A series of experiments are conducted to confirm the correctness of the presented approach and to demonstrate how to apply the developed solution in practical applications. The proposed method can be easily integrated with both newly created and existing solutions because it does not require the introduction of any modifications in the system structure and calibration process.     

Evaluation of quantization error in computer vision

Due to the important role that digitization error plays in the field of computer vision, a careful analysis of its impact on the computational approaches used in the field is necessary. The authors develop the mathematical tools for the computation of the average (or expected) error due to quantization. They can be used in estimating the actual error occurring in the implementation of a method. Also derived is the analytic expression for the probability density of error distribution of a function of an arbitrarily large number of independently quantized variables. The probability that the error of the function will be within a given range can thus be obtained accurately. The tools developed can be used in the analysis of the applicability of a given algorithm     

DeViouS: A distributed environment for computer vision

Computer vision, owing to the size and complexity of its tasks and its importance to industrial and economic growth, was selected as one of the grand challenge problems by the U.S. Federal High Performance Computing Program. Integration of vision operations is identified as a key element of the challenge. A system to integrate computer vision in a distributed environment is presented here. This system, called DeViouS, is based on the client/server model and runs in a heterogeneous environment of Unix workstations. Modern computing environments include large numbers of high-powered workstations connected by a very fast network. Many of these computers are idle most of the time. DeViouS takes advantage of this feature of computing environments to distribute the execution of vision tasks. Two primary goals of DeViouS are to provide a practical distributed system and a research environment for vision computing. DeViouS is based on a modular design that allows experimentation in various aspects of algorithm design, scheduling and network programming. It can make use of any existing computer vision packages with very minor changes to DeViouS. DeViouS has been tested in an environment of SUN and Digital workstations and has shown substantial improvements in speed over sequential computing with negligible overhead.     

Viz-A-Vis: toward visualizing video through computer vision

In the established procedural model of information visualization, the first operation is to transform raw data into data tables [1]. The transforms typically include abstractions that aggregate and segment relevant data and are usually defined by a human, user or programmer. The theme of this paper is that for video, data transforms should be supported by low level computer vision. High level reasoning still resides in the human analyst, while part of the low level perception is handled by the computer. To illustrate this approach, we present Viz-A-Vis, an overhead video capture and access system for activity analysis in natural settings over variable periods of time. Overhead video provides rich opportunities for long-term behavioral and occupancy analysis, but it poses considerable challenges. We present initial steps addressing two challenges. First, overhead video generates overwhelmingly large volumes of video impractical to analyze manually. Second, automatic video analysis remains an open problem for computer vision.     

Understanding minority costumes: a computer vision perspective

Multimedia Systems - It is an extremely interesting work to understand the minority costumes in computer vision and ethnology community. It explored some crucial clue for understanding minority...     

Query based visual analysis: spatio-temporal reasoning in computer vision

Two experimental systems for query-based visual analysis are described. The first simulates an image sequence of moving, dividing cells with simple rules and monitors significant visual events. The second processes single raw images of real cells. Both invoke appropriate processing using explicit knowledge to respond to user queries. It is proposed that this selectivity is an essential feature for any system to analyse raw image sequences of moving, dividing cells as the computational expense of allowing all possible processing to proceed is enormous. Processing as required by the query allows adaptive strategies (e.g. different resolutions and focal processing) to be utilized and gives an effective attentional control structure to the system.     

ISR: a database for symbolic processing in computer vision

ISR (international symbolic representation), a representation and management system for use at the intermediate (symbolic) level of vision, is described. ISR mediates access to intermediate-level vision data and forms an active interface to the higher-level inference processes that construct an image's interpretation. The system supports important types of data and operations and can be adapted to the changing needs of ongoing research. It provides a centralized data representation that supports integration of results from multiple avenues of research into the overall vision system. ISR's underlying computational paradigm is explained, database requirements for image interpretation are identified, the ISR data management system is described, and the use of ISR is discussed     

Deep reinforcement learning in computer vision: a comprehensive survey

Deep reinforcement learning augments the reinforcement learning framework and utilizes the powerful representation of deep neural networks. Recent works have demonstrated the remarkable successes of deep reinforcement learning in various domains including finance, medicine, healthcare, video games, robotics, and computer vision. In this work, we provide a detailed review of recent and state-of-the-art research advances of deep reinforcement learning in computer vision. We start with comprehending the theories of deep learning, reinforcement learning, and deep reinforcement learning. We then propose a categorization of deep reinforcement learning methodologies and discuss their advantages and limitations. In particular, we divide deep reinforcement learning into seven main categories according to their applications in computer vision, i.e. (i) landmark localization (ii) object detection; (iii) object tracking; (iv) registration on both 2D image and 3D image volumetric data (v) image segmentation; (vi) videos analysis; and (vii) other applications. Each of these categories is further analyzed with reinforcement learning techniques, network design, and performance. Moreover, we provide a comprehensive analysis of the existing publicly available datasets and examine source code availability. Finally, we present some open issues and discuss future research directions on deep reinforcement learning in computer vision.