Real-time Processing System For Single Interference Fringe Pattern

With the development of various research and application fields such as computer technology,optical technology,information technology and laser application technology,the requirement of high precision measurement has grown rapidly.Opticial interferometry techniques offer non-contact.full filed and high precision measurement for information such texture,shape,deviation,and has been widely used.The measurement results of optical interferometry techniques are fringe patterns,where phase demodulation is carried out to get the measured information.Sequential processing of fringe patterns cannot meet the real-time needs of object surface measurement in industrial production.Meanwhile,with the development of the GPU general computing performance and programmability,it has become a new hot spot in computer science and technology filed to use the powerful parallel computing ability of GPU and improve the processing speed,which provides a new idea to speed-up the interferogram processing.This paper intends to solve the conflicts between the large data processing requirement of interference fringe patterns and the long sequential computation of traditional CPU,which cannot meet the requirements of real-time processing in industry.Combined with the strong parallel computing ability of GPU,a real-time interference image processing system based on GPU is developed and the algorithm is optimized and improved.The system effectively improves the processing speed of interference image.There are three main steps in the design of the system.First is to denoise fringe patterns;then to extract the phase from fringe patterns;finally is to unwrap the wrapped phase.Through the phase analysis,we can obtain the hidden information on the surface of objects which cannot be seen with our bare eyes.