Bayesian Classification for Inspection of Industrial Products

AICT 2011

Bayesian, k-nearest neighbor, and Parzen window classifiers along with PCA and LDA methods, are effective tools in machine learning. In this work, a hybrid method is formed by the above mentioned methods. The aim is to achieve a successful, fast, and low computational classification. Performance of the new method is evaluated on five various kinds of datasets, from UCI machine learning datasets, including Breast Cancer, Iris, Glass, Yeast, and Wine. The experimental results indicate the superior performance of the proposed method in comparison with the previous works.

Journal of Intelligent Manufacturing, 2010

This paper presents a solution to a problem existing in the cork industry: cork stopper/disk classification according to their quality using a visual inspection system. Cork is a natural and heterogeneous (remarkable variability among different samples, being impossible to find two samples with the same morphological distribution in its defects) material; therefore, its automatic classification (seven quality classes exist) is very difficult. The solution proposed in this paper evaluates the following procedures: quality discriminatory features extraction and classifiers analysis. Each procedure focused on the study of aspects that could influence cork quality. Experiments show that the best results are obtained by system specific features: cork area occupied by defects (after thresholding), size of the biggest defect within the cork area (morphological operations), and the Laws TEMs E5L5TR, E5E5TR, S5S5TR, W5W5TR, all working on a Neuro-Fuzzy classifier. In conclusion, the results of this study represent an important contribution to improve quality control in the cork industry.

This paper addresses the problem of image categorization using local sensory information which is aggregated into global cortical-like representations of different image categories. Local information is adaptively extracted from an image database using IndependentComponent Analysis (ICA) whichprovides a set of localized, oriented, and band-pass filters selective to independent features of the different categories. Such local representations have been computationaly investigated by several researchers, and have also been experimentaly observed as characteristics of simple cells receptive fields in the primary visual cortex. However, very little work has been done on further use of these representations to provide more complex and global description of images. In this paper, we present an algorithm which uses the energy of a minimal set of filters to provide category-specific signatures which are shown to be strongly discriminant. Computer simulations are carried on an image d...

IECON Proceedings (Industrial Electronics Conference), 2010

This article presents a system developed for the industry of bottling beer. The system has to perform the inspection of various items in the final stage, meaning after the production phase where the bottles are already in the crate. The items to inspect are the following: whether the crate is correct (with the correct color), whether the crate is broken, whether the crate is fully populated, i.e., all bottles are present, to check for bottles without caps and whether the capsule is the correct one. The work uses techniques of computer vision for these verifications and also principal components analysis (PCAs) for the recognition of the capsules. This system is currently installed in the assembly line and the results indicate high efficiency and confidence in the obtained solution.

IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003

This paper is focused on the problems of feature selection and classification when classes are modeled by statistically independent features. We show that, under the assumption of class-conditional independence, the class separability measure of divergence is greatly simplified, becoming a sum of unidimensional divergences, providing a feature selection criterion where no exhaustive search is required. Since the hypothesis of independence is infrequently met in practice, we also provide a framework making use of classconditional Independent Component Analyzers where this assumption can be held on stronger grounds. Divergence and the Bayes decision scheme are adapted to this class-conditional representation. An algorithm that integrates the proposed representation, feature selection technique, and classifier is presented. Experiments on artificial, benchmark, and real-world data illustrate our technique and evaluate its performance.

… RECOGNITION AND IMAGE …, 2004

In this paper, a novel method for texture defect detection is presented. The method makes use of Independent Component Analysis (ICA) for feature extraction from the nonoverlapping subwindows of texture images and classifies a subwindow as defective or nondefective according to Euclidean distance between the feature obtained from average value of the features of a defect free sample and the feature obtained from one subwindow of a test image. The experimental results demonstrating the use of this method for visual inspection of textile products obtained from a real factory environment are also presented.

American Journal of Enology and Viticulture

Image-analysis techniques were applied to the surface of wine cork stoppers (tops and bodies) of the standard seven commercial quality classes to characterize their porosity. Canonical discriminant analysis (CDA) and stepwise discriminant analysis (SDA) were used to differentiate quality class and to identify the best features to select these classes. The accuracy of classification using CDA functions was on average greater than 50% for the seven commercial classes and was greater than 67% for a simplified three-grade classification. Based on the independent variables of the first CDA function determined by the stepwise method, a set of features was selected for use in decision rules for cork stopper classification: porosity coefficient and maximum pore dimensions (length and area) for bodies and porosity coefficient and number of pores for tops. Threshold limits for each feature were established for each quality class and a classification algorithm was applied. Results showed an overall match in class yield of 86% and better class homogeneity and separation. These are proposed as a foundation for future standardization of cork stopper classification based on image analysis and computerized vision systems selection of quantified features to ensure uniformity and transparency in trade while maintaining the overall economical feasibility in industrial processing.

Journal of Chemometrics, 2002

Principal component analysis (PCA) is a dimensionality reduction modeling technique that transforms a set of process variables by rotating their axes of representation. Maximum Likelihood PCA (MLPCA) is an extension that accounts for different noise contributions in each variable. Neither PCA nor its extensions utilize external information about the model or data such as the range or distribution of the underlying measurements. Such prior information can be extracted from measured data and can be used to greatly enhance the model accuracy. This paper develops a Bayesian PCA (BPCA) modeling algorithm that improves the accuracy of estimating the parameters and measurements by incorporating prior knowledge about the data and model. The proposed approach integrates modeling and feature extraction by simultaneously solving parameter estimation and data reconciliation optimization problems. Methods for estimating the prior parameters from available data are discussed. Furthermore, BPCA reduces to PCA or MLPCA when a uniform prior is used. Several examples illustrate the benefits of BPCA versus existing methods even when the measurements violate the assumptions about their distribution.

2006

This paper presents an automatic vision assortment method for cork classification. Cork is a natural material that is used to seal wine bottles due to its reliability and its chemical and mechanical properties. A colour image acquisition system is able to differentiate some cork defects. The developed and implemented system acquires images using a colour linear CCD camera, transmits that image through a fast acquisition card and processes that image to extract quantitative elements. This system can analyse and distinguish defects present on the surface of a cork-stopper: the size, form and position of the defects are analysed. Key-Words: Visual Inspection, Artificial Vision, Quality Control

… National Conference on …, 1988

This paper describes a Bayesian technique for unsupervised classification of data and its computer implementation, AutoClass. Given real valued or discrete data, AutoClass determines the most probable number of classes present in the data, the most probable descriptions of those ...