Browsing by Author "SU, CT"

Now showing 1 - 2 of 2

A COMPARISON OF STATISTICAL REGRESSION AND NEURAL-NETWORK METHODS IN MODELING MEASUREMENT ERRORS FOR COMPUTER VISION INSPECTION SYSTEMS
(1995-07-01)
CHANG, CA
;
SU, CT
;
工業工程與管理學系
;
Department of Industrial Engineering and Management
This paper compares measurement error models for computer vision inspection systems based on the statistical regression method and a neural network-based method. Experimental results demonstrate that both of the models can effectively correct the dimensional measurements of geometric features on a part profile. It also shows that the statistical regression method can perform excellent tasks when the functions for models are carefully selected through statistical testing procedures. On the other hand, varieties of neural network architectures all have good performance when training data are collected carefully. The explicit nonlinear relationship in neural network architectures is very effective in building a general mapping model without specifying the functional forms in advance. While statistical regression methods will continue to play important roles in model building tasks, the neural network-based method will be a very powerful alternative for precision measurement using computer vision systems.
NEURAL NETWORKS FOR PRECISE MEASUREMENT IN COMPUTER VISION SYSTEMS
(1995-11-01)
SU, CT
;
CHANG, CA
;
TIEN, FC
;
工業工程與管理學系
;
Department of Industrial Engineering and Management
Although computer vision systems have been successfully applied to some inspection tasks, they were generally not considered as precise measurement tools due to dimensional distortion and errors. This paper presents procedures to correct these errors for precise measurement. The first step is to formulate calibration models for image coordinate systems using neural networks. Then neural networks to model dimensional errors from the initial measurement are structured in a learning stage using standard parts. Finally these models are used to correct measurement errors in measurement tasks. These proposed procedures are implemented as an example.