Research On Auto-focus Methods Based On Digital Imaging Processing

Auto-focus is the key technology of imaging system in the process of imaging.With the development of computer technology,mechanical automation and image processing technology,auto-focus technology based on image processing are widely used in various types of imaging systems because of its simple system structure,fast focusing speed and high accuracy.The popularity of various small-sized imaging devices,such as digital cameras and smart camera phones,has increased the consumer demand for auto-focus technology.Both the quality and the speed of imaging are very important,how to achieve auto-focus quickly and accurately becomes the key point of this problem.However,the existing image processing based auto-focus technology still has some shortcomings and deficiencies,and there are still many problems to be solved.Therefore,it is of great significance to carry out research on auto-focus methods based on image processing in the field of optical imaging.This paper mainly focuses on the improvement of auto-fucos evaluation function,the selection of auto-focus window,the design of auto-focus searching scheme and the post-processing techniques of image,to meet the needs of consumers for auto-focus technology and solve the existing problems.Firstly,the optical imaging model and the basic principle of imaging system are analyzed,and the classification of the basic auto-focus methods are improved.We point out that the key point of auto-focus process based on image processing,which provides direction and basic ideas for our research on auto-focus thechnology.The research on auto-focus evaluation function is carried out.The classification and principle of existin evaluation function are introduced,and the property of ideal evaluation function are pointed out.An auto-focus evaluation function with adjustable coverage frequency based on absolut gradient value are proposed,which solves problem that the coverage of evaluation function is not wide enough when the range of focusing system is large and achieves the purpose of adjusting the coverage of evaluation function.We propose an evaluation function based on strong edge width histogram via block-weighted processing on image with edge information to further improve the performance of the algorithm.An evaluation function based on phase coherence is proposed,which utilize phase consistency to overcome the influence of changes in brightness,contrast and low signal-to-noise ratio on image quality evaluation.The necessity and property of focusing window selection are discussed,and the existing focusing window selection method are introduced.We found out that there is a phenomenon of scene drift during auto-focus process and studied the focus window tracking slection strategy.Research on focus window selection algorithm for multiple depth scenes,and on this basis,the background blur algorithm for the in-focus image is studied to realize the artistic effect of large aperture for the imaging system with volume limite.Carried out research on the design of auto-focus searching scheme.Principle of several auto-focus searching strategies,whose advantages and disadvantages aare introduced.In order to meet the real-time and accuracy requirments of consumer,an expression of defocus model based on the reciprocal curve is deduced by studing the optical defocus model.And then,an optical defocus fitting model-based auto-focus scheme is proposed,and its feasibility is verified by experiments.Researches on auto-focus and post-processing techniques are conducted for the special scenes,such as night environment.For the auto-focus process affected by the bright areas such as lights,an evaluation function which removes the influence of light source is proposed.A low-light image enhancement algorithm based on dark channel principle is proposed to enhance the image and recover details of the dark region.In addition,a denoise algorithm based on noise estimation is proposed to improve the image quality and achieve satisfactory imageing results.