Optimization Of Measurement Results Of Lens Modulation Transfer Function Based On Artificial Intelligence Algorithm

With the development of modern optics,it’s known that the optical system can be regarded as a frequency filter;this means Image quality can be evaluate by comparing the frequency spectrum of object and image,and the transfer properties of different frequencies by optical system is optical transfer function.The modulation transfer function(MTF)of the optical transfer function is the key to determine the image quality of optical component,this new method is not only accepted by photographers,but it also applied in optical measurement,industrial processing and so on.MTF is closely related to focus position,but it’s difficult to obtain the best focus position by theoretical solution considering the instrument error and experimental error,and manually adjustment can lead more errors.In recent years,with the development of artificial intelligent technology,in the aspect of optimization method,some optimization methods based on artificial intelligence are proposed,which are different from traditional optimization methods.Among them,neural networks and support vector machines have been more and more used for function approximation and Simulation of black box model because of their advantage of approaching non-linear data.On this basis,the main tasks in this paper were listed as follows:(1)The physical meaning and measurement method of results of MTF is discussed.Furthermore,the MTF optical testing system and platform are established,and the key factors of results of MTF are explored.(2)A optimization algorithm of results of MTF is proposed based on wavelet neural network,the best focus position is obtained.Moreover,the calculated results of MTF of WNN and manually adjustment is compared and analyzed.(3)A optimization algorithm of results of MTF is proposed based on support vector machine,the best focus position is obtained.Moreover,the calculated results of MTF of SVM and manually adjustment is compared and analyzed.The main conclusions were listed as follows:(1)Results of MTF can be used to evaluate the image quality of optical components,and the results of MTF is in direct proportion to the distance of CMOS camera and the best focus position.(2)The calculated results of MTF of WNN is better than manually adjustment,and WNN method can obtain a better position of CMOS camera.(3)The calculated results of MTF of SVM is better than manually adjustment,and SVM method can obtain a better position of CMOS camera.