Research On Microfocusing Algorithm Based On Depth From Focus

Image is an important carrier to reflect the objective world and clear image is helpful to people's research and exploration.As one of key technologies in modern imaging systems,auto-focusing is used to assist imaging systems to capture clear images and widely used in various fields.It's also beneficial to improve the accuracy of fiber identification to obtain clear fiber image with auto-focusing technology in fiber detection.At present,the depth from focus method based on image processing is mainly used in the automatic microscope focusing system.The depth from focus method is mainly composed of three steps: focus evaluation function calculation,focus window selection and search strategy.This paper carries on the thorough research on the depth from focus method,and the work of this method is arranged as follows:Firstly,this paper established out-of-focus model of image system and brought point spread function and optical transfer function.Also analyzeed the causes of image blurring and laid a theoretical foundation for automatic focus research.Secondly,the focus evaluation function was studied.Based on the principle of focused evaluation,the construction basis of evaluation function was explained.According to the features of fractional differentiation which can greatly improve the information of image edge and texture details,fractional differentiation was introduced into the focus evaluation.In order to evaluate the quality of the fiber image more effectively,1?2 order differential operators were constructed,which combined with the second order integer differentiation to form the evaluation function.The purpose is to strengthen the detection of intermediate frequency information while detecting the high-frequency information,and integrate the high-frequency and intermediate frequency information of the image as the evaluation result of the image clarity.The simulation and comparison experiments with various evaluation functions were carried out,and the effectiveness of the proposed evaluation functions was proved by qualitative and quantitative analysis.Thirdly,how to construct the focusing window was studied.Textile fibers are randomly distributed in microscope field.The traditional focusing window select method can not capture the target subject accurately.Based on the analysis and comparison of several commonly used focusing windows,a method of extracting windows from chaotic artificial fish swarm was proposed.The global convergence of fish swarm algorithm was improved by adding chaos search algorithm to traditional fish swarm algorithm.The search step and visualfield were changed at real-time to improve its search accuracy.The entropy value of the image in the window was calculated by using the gray symbiotic moment as the comparison basis.The result shows that the window acquired by the improved fish swarm algorithm contains more target contents.The comparison between the two methods of fish shoal window extraction and the traditional method of window extraction shows that this paper's focusing window select method is more adaptive.Finally,the microscope automatic focusing experiment platform was built by the combination of upper computer and lower computer and the hardware and software of the platform were designed.The proposed focusing scheme was tested on the self-developed experimental platform.The focusing method proposed in this paper has good effect on focusing textile fiber and get ready for automatic testing of textile fibers.