White Light Interference Measurement Signal Processing Based On Generalized Gaussian Distribution And FPGA Implementation

Scanning white light interferometry to measure the surface of the surface micro-three-dimensional morphology in the detection of the fracture surface has a potential application prospects.In the inspection of the surface of the fracture found that there are a lot of cracks and bumps,these irregular morphology will fracture the surface of the original shape of the recovery of serious interference.In this paper,we use wavelet analysis based on generalized Gaussian distribution model to decompose and collect the collected signals.But the complex operation will increase the system power consumption,increase the volume of the system,reduce the flexibility of the entire system,and the system can not be real-time signal acquisition and transmission and processing.So the need for the entire hardware resources to design and optimization,to enhance the flexibility of the entire system.This paper uses FPGA to design,and the entire system of timing control and logic operations to it for processing.It can improve the efficiency of the entire system,reduce its power consumption,to ensure the orderly and stable operation of the entire system.At the same time the development of the system is also applicable to the object surface quality testing,object recognition,cultural relics scanning,robot navigation,medical scanning and other scientific research.In this paper,the generalized Gaussian distribution model is used to calculate the wavelet coefficients of the white light interference signal in the same layer.By estimating the shape parameters and scale parameters corresponding to the wavelet coefficients after each wavelet transform.The article calculates the thresholds corresponding to each set of parameters and then the resulting thresholds are substituted into the soft and hard threshold function or the further improved threshold function to filter the wavelet coefficients.In this paper,we compare the denoising effects of different threshold functions when the theory is modeled and simulated.We use the control variable method to compare the threshold acquisition method horizontally and vertically: in the horizontal threshold comparison,we establish the generalized Gaussian distribution,the Gaussian distribution and the Laplace distribution model respectively,and then get the threshold to compare the denoising effect.We compare the thresholds obtained by the generalized Gaussian distributionwith the VisuShrink Threshold,the SureShrink Threshold,the HeurShrink Threshold,and the MinmaxShrink Threshold.Finally,this paper evaluates the effect of different threshold processing methods and models on the effect of interfering signal processing.After the theoretical modeling is completed,the hardware denoising module is designed by 5/3 lifting wavelet.Finally,a general scheme based on the general framework of FPGA wavelet threshold denoising system and various sub-modules is designed:Including two-dimensional wavelet denoising module,SDRAM fast storage master and interface module,VGA driver interface module and CCD data acquisition module,and the overall module and the various system modules were simulated and verified.MATLAB theoretical analysis:In this paper,the generalized Gaussian distribution model is used to calculate the wavelet coefficients of the white light interference signals in the same layer,and the optimal thresholds for the current interferometric signals can be estimated.The wavelet coefficients of the decomposed wavelet are filtered by lifting the threshold function of the compromise algorithm,which has better denoising effect compared with other methods.Hardware logic validation indicates that: The article quickly achieves the multiplication and division of the system by the shift operation,and the principle of time division multiplexing coordination of data storage,data processing,data display the relationship between the three.The system can be real-time image noise reduction processing,its can also be transplanted to other on the wavelet denoising image processing system,with a certain theoretical and practical value.