Bruce G. Batchelor

Machine vision for the inspection of natural products

List of Contributors 1. Like Two Peas in a Pod B.G. Batchelor Editorial Introduction 1.1 Advantages of Being Able to See 1.2 Machine Vision 1.2.1 Model for Machine Vision Systems 1.2.2 Applications Classified by Task 1.2.3 Other Applications of Machine Vision 1.2.4 Machine Vision Is Not Natural 1.3 Product Variability 1.3.1 Linear Dimensions 1.3.2 Shape 1.3.3 Why Physical Tolerances Matter 1.3.4 Flexible and Articulated Objects 1.3.5 Soft and Semi-fluid Objects 1.3.6 Colour Variations 1.3.7 Transient Phenomena 1.3.8 Very Complex Objects 1.3.9 Uncooperative Objects 1.3.10 Texture 1.4 Systems Issues 1.5 References 2. Basic Machine Vision Techniques B.G. Batchelor and P.F. Whelan Editorial Introduction 2.1 Representation of Images 2.2 Elementary Image Processing Functions 2.2.1 Monadic Point-by-point Operators 2.2.2 Dyadic Point-by-point Operators 2.2.3 Local Operators 2.2.4 Linear Local Operators 2.2.5 Non-linear Local Operators 2.2.6 N-tuple Operators 2.2.7 Edge Effects 2.2.8 Intensity Histogram [hpi, hgi, he, hgc} 2.3 Binary Images 2.3.1 Measurements on Binary Images 2.3.2 Shape Descriptors 2.4 Binary Mathematical Morphology 2.4.1 Opening and Closing Operations 2.4.2 Structuring Element Decomposition 2.5 Grey-scale Morphology 2.6 Global Image Transforms 2.6.1 Hough Transform 2.6.2 Two-dimensional Discrete Fourier Transform 2.7 Texture Analysis 2.7.1 Statistical Approaches 2.7.2 Co-occurrence Matrix Approach 2.7.3 Structural Approaches 2.7.4 Morphological Texture Analysis 2.8 Implementation Considerations 2.8.1 Morphological System Implementation 2.9 Commercial Devices 2.9.1 Plug-in Boards: Frame-grabbers 2.9.2 Plug-in Boards: Dedicated Function 2.9.3 Self-contained Systems 2.9.4 Turn-key Systems 2.9.5 Software 2.10 Further Remarks 2.11References 3. Intelligent Image Processing B.G. Batchelor Editorial Introduction 3.1 Why We Need Intelligence 3.2 Pattern Recognition 3.2.1 Similarity and Distance 3.2.2 Compactness Hypothesis 3.2.3 Pattern Recognition Models 3.3 Rule-based Systems 3.3.1 How Rules are Used 3.3.2 Combining Rules and Image Processing 3.4 Colour Recognition 3.4.1 RGB Representation 3.4.2 Pattern Recognition 3.4.3 Programmable Colour Filter 3.4.4 Colour Triangle 3.5 Methods and Applications 3.5.1 Human Artifacts 3.5.2 Plants 3.5.3 Semi-processed Natural Products 3.5.4 Food Products 3.6 Concluding Remarks 3.7 References 4. Using Natural Phenomena to Aid Food Produce Inspection G. Long Editorial Introduction 4.1 Introduction 4.2 Techniques to Exploit Natural Phenomena 4.3 Potato Sizing and Inspection 4.4 Stone Detection in Soft Fruit Using Auto-fluorescence 4.5 Brazil Nut Inspection 4.6 Intact Egg Inspection 4.7 Wafer Sizing 4.8 Enrobed Chocolates 4.9 Conclusion 4.10 References 5. Colour Sorting in the Food Industry S.C. Bee and M.J. Honeywood Editorial Introduction 5.1 Introduction 5.2 The Optical Sorting Machine 5.2.1 The Feed System 5.2.2 The Optical System 5.2.3 The Ejection System 5.2.4 The Image Processing Algorithms 5.3 Assessment of Objects for Colour Sorting 5.3.1 Spectrophotometry 5.3.2 Monochromatic Sorting 5.3.3 Bichromatic Sorting 5.3.4 Dual Monochromatic Sorting 5.3.5 Trichromatic Sorting 5.3.6 Fluorescence Techniques 5.3.7 Infrared Techniques 5.3.8 Optical Sorting with Lasers 5.4 The Optical Inspection System 5.4.1 Illumination 5.4.2 Background and Aperture 5.4.3 Optical Filters 5.4.4 Detectors 5.5 The Sorting System 5.5.1 Feed 5.5.2 Ejection 5.5.3 Cleaning and Dust Extraction 5.5.4 The Electronic Processing System 5.6 The Lim

Software implementation of the SKIPSM paradigm under PIP

SKIPSM (separated-kernel image processing using finite state machines) is a technique for implementing large-kernel binary- morphology operators and many other operations. While earlier papers on SKIPSM concentrated mainly on implementations using pipelined hardware, there is considerable scope for achieving major speed improvements in software systems. Using identical control software, one-pass binary erosion and dilation structuring elements (SEs) ranging from the trivial (3 by 3) to the gigantic (51 by 51, or even larger), are readily available. Processing speed is independent of the size of the SE, making the SKIPSM approach practical for work with very large SEs on ordinary desktop computers. PIP (prolog image processing) is an interactive machine vision prototyping environment developed at the University of Wales Cardiff. It consists of a large number of image processing operators embedded within the standard AI language Prolog. This paper describes the SKIPSM implementation of binary morphology operators within PIP. A large set of binary erosion and dilation operations (circles, squares, diamonds, octagons, etc.) is available to the user through a command-line driven dialogue, via pull-down menus, or incorporated into standard (Prolog) programs. Little has been done thus far to optimize speed on this first software implementation of SKIPSM. Nevertheless, the results are impressive. The paper describes sample applications and presents timing figures. Readers have the opportunity to try out these operations on demonstration software written by the University of Wales, or via their WWW home page at http://bruce.cs.cf.ac.uk/bruce/index.html .

Interactive Image Processing for Machine Vision

Machine vision systems offer great potential in a large number of areas of manufacturing industry and are used principally for Automated Visual Inspection and Robot Vision. This publication presents the state of the art in image processing. It discusses techniques which have been developed for designing machines for use in industrial inspection and robot control, putting the emphasis on software and algorithms. A comprehensive set of image processing subroutines, which together form the basic vocabulary for the versatile image processing language IIPL, is presented. This language has proved to be extremely effective, working as a design tool, in solving numerous practical inspection problems. The merging of this language with Prolog provides an even more powerful facility which retains the benefits of human and machine intelligence. The authors bring together the practical experience and the picture material from a leading industrial research laboratory and the mathematical foundations necessary to understand and apply concepts in image processing. Interactive Image Processing is a self-contained reference book that can also be used in graduate level courses in electrical engineering, computer science and physics.

Using Dynamic Condor-Based Services for Classifying Schizophrenia in Diffusion Tensor Images

Diffusion tensor imaging (DTI) provides insight into the white matter of the human brain, which is affected by schizophrenia. By comparing a patient group to a control group, the DTI-images are on average expected to be different for white matter regions. Principal component analysis (PCA) and linear discriminant analysis (LDA) are used to classify the groups. In this work, the number of principal components is optimised for obtaining the minimal classification error. A robust estimate of this error is computed in a cross-validation framework, using different compositions of the data into a training and a testing set Previously, sequential runs were performed in MATLAB, resulting in long execution times. In this paper we describe an experiment where this application was run on a grid with minimal modifications and user effort. We have adopted a service-based approach that autonomously launches image analysis services onto a campus-wide Condor pool comprising of volunteer resources. This allows high throughput analysis of our data in a dynamic resource pool. The challenge in adopting such an approach comes from the nature of the resources, which change randomly with time and thus require fault tolerance. Through this approach we have reduced the computation time of each dataset from 90 minutes to less than 10. A minimal classification error of 22% was obtained, using 15 principal components.

Dynamic Condor-based Services for Distributed Image Analysis

Interactive image processing is an important requirement in many industrial applications, such as the inspection of industrial parts within a manufacturing environment, or the processing of images from surveillance cameras. Being able to achieve this quickly and accurately is often essential for the success of such industrial applications. A service-based approach that autonomously launches Image Analysis Services (accessible through a Central Service Manager) onto spare network resources through a Condor system is presented. This allows high throughput analysis of these images in a dynamic resource pool. The Central Service Manager reacts to new tasks submitted to the Image Analysis Services and is able to add new service instances to manage these tasks dynamically. Each service instance here corresponds to a computational resource that is able to execute image processing algorithms. New service instances may be requested by the Central Service Manager from the Condor system, based on the number of tasks that need to be processed. This enables entire image repositories to be acted upon interactively and in parallel, as opposed to the analysis of single images individually. The approach is demonstrated through a campus-wide test bed utilising a Condor system with 90 machines.

Machine vision for industrial applications

Vision is critical for the survival of almost all species of the higher animals, including fish, amphibians, reptiles, birds, and mammals. In addition, many lower animal phyla, including insects, arachnids, Crustacea, and molluscs, possess well-developed optical sensors for locating food, shelter, a mate, or a potential predator. Even some unicellular organisms are photosensitive, even though they do not have organs that can form high-resolution images. Vision bestows great advantages on an organism. Looking for food, rather than chasing it blindly, is very efficient in terms of the energy expended. Animals that look into a crevice in a rock to check that it is safe to go in are more likely to survive than those that do not do so. Animals that use vision to identify a suitable mate are more likely to find a fit, healthy one than those that ignore the appearance of a possible partner. Animals that can see a predator approaching are more likely to be able to escape capture than those that cannot. Compared with other sensing methods based on smell, sound, and vibration, vision offers far greater sensitivity and resolution for discriminating the four essentials for survival listed above.

Agent-based recognition of facial expressions

Description of a system to detect facial expressions using an agent-based approach is presented. The system utilizes interaction between Matlab-based image filters and a JADE-based agent implementation. The system is demonstrated using a feature recognition example. The system however has a much wider applicability, especially as Matlab is used extensively in other scientific/numerical computing applications.

Vision Systems on the Internet

Hitherto, it has been assumed that an industrial Machine Vision systems is constructed as an integrated unit, with the camera, image processing unit and control/display console being located close to one another and to the object/scene being inspected. For several reasons, it may be helpful to separate them, so that only the camera and its associated lighting units are located on the factory floor, while other equipment, such as computers and user terminals, is located elsewhere, out of harms way. We describe three systems that allow multiple cameras and several separate image processing engines, to be controlled remotely from a single intelligent device, or a user working via a standard web browser. The paper describes and compares several different approaches to building such a system. Links to the networked vision systems mentioned here are provided in an accompanying web site.

Networked vision system using a prolog controller

Prolog offers a very different style of programming compared to conventional languages; it can define object properties and abstract relationships in a way that Java, C, C++, etc. find awkward. In an accompanying paper, the authors describe how a distributed web-based vision systems can be built using elements that may even be located on different continents. One particular system of this general type is described here. The top-level controller is a Prolog program, which operates one, or more, image processing engines. This type of function is natural to Prolog, since it is able to reason logically using symbolic (non-numeric) data. Although Prolog is not suitable for programming image processing functions directly, it is ideal for analysing the results derived by an image processor. This article describes the implementation of two systems, in which a Prolog program controls several image processing engines, a simple robot, a pneumatic pick-and-place arm), LED illumination modules and a various mains-powered devices.

2D packing using the Myriad framework

Myriad is a framework for building networked and distributed vision systems and is described in a companion paper in this conference. Myriad allows the components of a multi-camera, multi-user vision system (web-cameras, image processing engines, intelligent device controllers, databases and the user interface terminals) to be interconnected and operated together, even if they are physically separated by many hundreds, or thousands, of kilometres. This is achieved by operating them as Internet services. The principal objective in this article is to illustrate the simplicity of harmonising visual control with an existing system using Myriad. However, packing of 2-dimensional blob-like objects is of considerable commercial importance in some industries and involves robotic handling and/or cutting. The shapes to be packed may be cut from sheet metal, glass, cloth, leather, wood, card, paper, composite board, or flat food materials. In addition, many 3D packing applications can realistically be tackled only by regarding them as multi-layer 2D applications. Using Myriad to perform 2D packing, a set of blob-like input objects (shapes) can be digitised using a standard camera (e.g. a webcam). The resulting digital images are then analysed, using a separate processing engine, perhaps located on a different continent. The packing is planned by another processing system, perhaps on a third continent. Finally, the assembly is performed using a robot, usually but not necessarily, located close to the camera.