Noncontact measurement using line-scan cameras: Analysis of positioning error

A multiprocessor camera controller has been designed and developed for real-time operation of high-resolution industrial CCD (charge coupled device) line-scan cameras. A three-camera system is capable of measuring length, width, height, and volume of steel slabs. Data captured by one camera are made available to other camera processors. A computation scheme is developed to correlate information for accurate cooperative measurement. There are two major sources of measurement errors. Digitizing error has been examined elsewhere, and a 0.1 subpixel accuracy is achievable by appropriate processing. The authors consider positioning errors with emphasis on camera positioning. The cooperative measurement and computation scheme measures object translation and compensates its effect to a certain degree. It is shown that, with calibration, measurement errors caused by camera positioning can be kept error caused by camera positioning can be kept within 0.2%. >

[1]  Chih-Shing Ho Precision of Digital Vision Systems , 1983, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  Tod S. Levitt,et al.  A Real-Time Automated Visual Inspection System for Hot Steel Slabs , 1983, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[3]  T. Y. Young,et al.  Subpixel edge detection and estimation with a microprocessor-controlled line scan camera , 1988 .

[4]  Tzay Y. Young,et al.  Subpixel Edge Detection And Estimation With A Line Scan Camera , 1987, Other Conferences.