Team
Presentation Slides
A report of this project is available up on request. Please contact us.
In the proposed research, we will focus on the development of a methodological framework that makes possible the automated development of AVI inspection algorithms for SMD components. Under this notion the resulting development framework will minimize the intervention of human developers to obtain inspection algorithms that meet minimal performance measures in terms of component discrimination and sensitivity to changes in environmental variables.
NSF Grant DMI-0300361
This main project includes the following sub-projects:
Automated Feature Selection for Visual Inspection Systems
The electronics assembly industry has faced the problem of rapid introduction and retirement of electronic products. Therefore, a system is required that automatically, and significantly shortens the time and that is economically feasible to develop inspection algorithms for new components or modifications in the actual products. The general goal of this research is to develop a self-training classifier for the inspection of Surface Mounted Devices (SMD) components. During the training phase of the classifier, feature selection (also referred to as variable selection) is necessary to reduce the computational cost and minimizes the inspection errors of the systems in the inspection phase. In particular, this paper explores the use of Multivariate Stepwise Discriminant Analysis techniques such as; Wilks’ Lamba, Unexplained Variance, Mahanalobis Distance, Smallest Distance, and Rao’s V in order to expedite the feature selection process.
NSF Grant DMI-0300361
Class Separability and Outlier Elimination for Quadratic Classification Vector
The Automated Visual Inspection (AVI) of Surface Mounted Devices (SMD) requires the correct classification of an image as either “component present” or “component absent.” The inspection system must allow the classification to be fast and reliable, while also assuring that the training of the classifier is simple and not time consuming. But a principal problem is how determine if the information that is in the classifier is useful to gives a good discrimination between the population, also called class separability. The outlier’s elimination of the population can be used as a technique to improve the discrimination between the classes.
NSF Grant DMI-0300361
Development of a Feature Selection Methodology for Automated Visual Inspection Systems
This study explores the use of multivariate discriminant procedures as an approach to the selection of feature subsets for a given SMD component. Furthermore, this research involves the careful evaluation of the behavior of a set of known features for the automated visual inspection system in place at Electronics Assembly Laboratory in Arizona State University in order to lead to the design of a framework for a heuristic that will expedite the feature selection process.
NSF Grant DMI-0300361
The objective of this proposal is the development of tools to significantly shorten the time required to develop inspection algorithms for new components. In particular, we propose to develop the methodology for the automatic generation of inspection routines of new SMD components introduced to a PCB assembly line. The envisioned methodology will make possible an AVI system that develops and optimizes its own inspection algorithms for new components.
In the proposed research, we will focus on the development of a methodological framework that makes possible the automated development of AVI inspection algorithms for SMD components. Under this notion the resulting development framework will minimize the intervention of human developers to obtain inspection algorithms that meet minimal performance measures in terms of component discrimination and sensitivity to changes in environmental variables.
NSF Grant DMI-0300361
This main project includes the following sub-projects:
Automated Feature Selection for Visual Inspection Systems
The electronics assembly industry has faced the problem of rapid introduction and retirement of electronic products. Therefore, a system is required that automatically, and significantly shortens the time and that is economically feasible to develop inspection algorithms for new components or modifications in the actual products. The general goal of this research is to develop a self-training classifier for the inspection of Surface Mounted Devices (SMD) components. During the training phase of the classifier, feature selection (also referred to as variable selection) is necessary to reduce the computational cost and minimizes the inspection errors of the systems in the inspection phase. In particular, this paper explores the use of Multivariate Stepwise Discriminant Analysis techniques such as; Wilks’ Lamba, Unexplained Variance, Mahanalobis Distance, Smallest Distance, and Rao’s V in order to expedite the feature selection process.
NSF Grant DMI-0300361
Class Separability and Outlier Elimination for Quadratic Classification Vector
The Automated Visual Inspection (AVI) of Surface Mounted Devices (SMD) requires the correct classification of an image as either “component present” or “component absent.” The inspection system must allow the classification to be fast and reliable, while also assuring that the training of the classifier is simple and not time consuming. But a principal problem is how determine if the information that is in the classifier is useful to gives a good discrimination between the population, also called class separability. The outlier’s elimination of the population can be used as a technique to improve the discrimination between the classes.
NSF Grant DMI-0300361
Development of a Feature Selection Methodology for Automated Visual Inspection Systems
This study explores the use of multivariate discriminant procedures as an approach to the selection of feature subsets for a given SMD component. Furthermore, this research involves the careful evaluation of the behavior of a set of known features for the automated visual inspection system in place at Electronics Assembly Laboratory in Arizona State University in order to lead to the design of a framework for a heuristic that will expedite the feature selection process.
NSF Grant DMI-0300361