| Optical interference measurement is a measurement technique combining with lasertechnology, video technology and computer image processing technology, widely used in thefiled of finite height measurement,object’s3D surface,refractive index of optical medium andtransparent object’s thickness. As the core of optical measurement techniques, opticalinterference image processing algorithms have got much attention. In this paper, the researchwork is developed with image processing algorithm,focused on some critical algorithms,such as filter, optical interference fringe center extracted and sub-pixel positioning.Firstly, pointed out their advantages and disadvantages through analysing the traditionalimage filtering method and partial differential equation filtering. And discussed opticalinterference fringe’s direction and density, quantified these two features. Based on the features,improved the anisotropic diffusion partial differential equation, which can protect the detailsand remove the noise simultaneously. Secondly, on the stripe center line extraction algorithm,used the MBO (Merriman-Bence-Osher) based on the heat conduction equation. In order toget better binary effect, the thresholds of MBO algorithm is improved, use the Bernsen localthreshold method instead of the global threshold. The improved method can effectivelybalance the light and improved the stripe center line extraction accuracy. Through the analysisof common sub-pixel algorithms, eventually use sub-pixel located algorithm based on Zernikemoments to position the centre of the circular fringes. As the experimental data shown, boththe positioning accuracy and cost-time can meet the actual needs. |
