Research On 3D Measurement Of Structural Wire Based On Computer Vision

Structural wire is a general term for steel wire used for cutting in watch industry and solar photovoltaic industry,which is also known as sawing wire.As an inexhaustible and clean energy,solar energy has been used more and more widely.Crystalline silicon is the main approach to use solar energy,and most solar silicon wafers are cut using structural wire.The shape parameters of the structural wire have an important influence on the surface quality of the silicon wafer.Structural wire is made from straight steel wire using special gears to form two orthogonal sine waves on horizontal and vertical dimensions.Amplitude of waves on the two dimensions are different,and these two waves are called big wave and small wave respectively.Structural wire shape parameters mainly refer to wave height and wave length of big wave and small wave.However,at present,there is no widely accepted measurement method or uniform measurement standard in field of structural wire shape parameters measurement.Structural wire manufacturers keep their technology as top secret.Quality evaluation of produced structural wire only rely on feedback of downstream business without scientific measurement standards.According to situation of structural wire industry,this paper uses mathematical equation to describe the structural wire,which is the vector superposition of two orthogonal sine waves.According to mathematical model of structural wire,this paper does researches on the projection images of structural wire.On condition of enough image observation length,while the rotation angle θ = n × 90°(n ∈ N),the projection images are standard sine waves.While θ takes other values,the projection images are cyclical curves.This paper introduces serveral separation algorithms of big wave and small wave based on research on the projection images.While observation length is longer than a period,frequency domain separation algorithm can achieve excellent result,because the spectrum in frequency domain is bimodal.This paper introduces the sine degree evaluation algorithm to seperate big wave and small wave while the observation length is not enough.This paper introduces the structural wire auto measurement system based on theoretical analysis above.The whole system is designed in view of computer vision principle,which is combined by two parts– measuring instrument and PC.Measuring instrument part completes jobs including fixing structural wire,rotation and image acquisition.PC part is responsible for image processing and recognition,and calculate wave height and wave length parameters.After the system development work is completed,serveral experiments are conducted.From the experiment data,big wave and small wave can be seperate exactly.Repeatability of wave height measurement less than ±1μm,and single measurement process less than 30 s,which is a substantial increase in structural wire measurement field.