environment based on OpenMP parallel technology

 

 

ZHU Jin, WU Kaihua

 

(College of Life Information Science & Instrument Engineering,

Hangzhou Dianzi University, Hangzhou 310018, China)

 

 

Abstract: During the online detection of wheel set wear based on optoelectronic image, the speed of wheel set image processing has significant impact to the realtime of online detection. Aiming at the problem that the wheel set image processing speed can not meet the requirement of high speed wheel set online detection, the calculation advantage of multicore processors was used and the wheel set image processing method based on OpenMP parallel technology was studied. Through the analysis of the characteristics of wheel set image, according to the different type of wheel set image, different segmentation method was used and the influence of wheel set image segmentation step to the postprocessing was reduced; the programming mode of multi image parallel processing was proposed and the parallel optimization scheme was designed, the OpenMP was used to realize the parallel optimization of the processing of wheel set image. Then the image processing effect of OpenMP four threads and single thread was compared to verify the processing acceleration effect of wheel set image through the experiment. The research results indicated that, based on multicore, the parallel optimization of wheel set image processing by OpenMP has improved the speed more than 2 times and met the requirement of wheel set online detection under 80km/h speed.

 

Key words: multicore; wheel set image; parallel optimization; OpenMP