神經回路網을 利用한 三次元形狀데이터分類에 관한 硏究

Abstract

The 3D measurement system which improves the touch probe 3D measurement defect, which calculates feet data and parameters. This system uses eight CCD(charge-coupled device) cameras with which it is equipped at the top and bottom, right and left sides, and four lasers which are also attached to both sides and upper and lower sides. These compound and shape numerical value data which are length, height, width and curved surface etc, in three dimensions. Also, this system extracts optimized data using BP(back propagation) algorithm of neural networks.

The whole system consists of a data acquisition part, an image data processing part, and a data output part. The data acquisition part consists of a measurement frame, eight CCD cameras and four lasers, and a control device for control. The image data processing part extracts feature data from feet. Finally the data output part confirms classified result data.

Fifteen parameters reading were acquired from the 3D measurement system improved data classification quality which was applied to the neural networks algorithm more than data classification quality which was not applied to the neural networks algorithm in classification process for optimized data.