Title :
Defects Detection in Web Inspection Based on DM642 DSP
Author :
Wang, Lei ; Ding, Yang ; Han, Zhijun
Author_Institution :
Key Lab. of Inf. Coding & Transm., Southwest Jiaotong Univ., Chengdu, China
Abstract :
In this paper, a novel approach for defects detection in web inspection based on TMS320DM642 DSP development board is proposed. The whole commercial web inspection system can be divided into three major function units: image capture, real-time processing of line-scanned images and the networking transmission. An innovative line-analysis based threshold detection algorithm is presented. The experimental results corroborate that this real-time method implements efficiently in the limited resources on a single DSP with high detection accuracy and robustness to the linear grayscale change. Though the analysis, we can see the total process ability of this algorithm finally achieves 22.51 M/s, which is nearly five to ten times as large as the other algorithms such as SOM based on statistics running on general processor without any obvious decline of the detection accuracy.
Keywords :
digital signal processing chips; image processing; inspection; paper industry; production engineering computing; TMS320DM642 DSP development board; Web defects detection; Web inspection; digital signal processing; image capture unit; image scanning unit; line analysis; networking transmission unit; threshold detection algorithm; Application software; Change detection algorithms; Circuits; Digital signal processing; Ethernet networks; Hardware; Image processing; Inspection; Real time systems; Signal processing algorithms; DM642; accuracy; real-time; scanning line;
Conference_Titel :
Computer Science-Technology and Applications, 2009. IFCSTA '09. International Forum on
Print_ISBN :
978-0-7695-3930-0
Electronic_ISBN :
978-1-4244-5423-5
DOI :
10.1109/IFCSTA.2009.106