Study On Auto-focusing Technology Based On Image Processing

Since the middle of the twentieth century,photoelectric imaging technology has been developing rapidly,and is now widely used in various fields such as photography,traffic monitoring,military detection,medical imaging and astronomical observation.The related researches of auto-focusing technology,one of the core technologies of digital imaging systems,has been gaining popularity among scientists and researchersIn the early stages,cameras mainly worked by manual focusing.The inefficient and outmoded focusing method,however,has been quickly replaced by automatic focusing as a result of developments of industrial technology and optical theories.Currently,the modern auto-focus systems based on image processing have become predominant in production and application.These systems use a photoelectric sensor to receive front-end lens imaging and convert them into electrical signals.Then the computer evaluates the focus degree of the digital image after the analog-to-digital conversion,and finally,drives the motor to achieve search and positioning by lens to complete the imaging focusing.Modern auto-focusing systems are mostly built in accordance with the Depth-From-Focus and the Depth-From-Defocus method.The Depth-From-Focus method,more widely employed,is mainly composed of three parts:focus window selection,focus evaluation function and focus search strategy.The focus window is designed to find the foreground target of the whole image.While reducing the workload of the evaluation function,a reasonable focus area can also effectively reduce the interference of noise and other factors of the background region.As the most important step of the auto-focus process,the focus evaluation function uses a specific algorithm to measure the sharpness of the imaging results at different focal lengths to provide a reference standard for lens adjustment.Finally,the focus search strategy is aimed to plan the stroke of the stepper motor to accurately achieve focus positioning and improve focus efficiency.The research content of this article focuses on the above three aspectsThis paper proposes improvements and innovations in the following aspects:(1)On account of the traditional focus evaluation functions' incompetence to balance accuracy and stability,and its gap in anti-noise ability,a sharpness evaluation function based on dynamic segmentation and local maximum gradient is proposed.This method combines the two-dimensional Otsu algorithm and the improved maximum gradient method to binarize the image to enhance the edge detailed contrast between the defocused image and the focused image,and then build a new sharpness evaluation function based on gradient variance and the variation characteristics of gradient non-zero solution.The simulation experiments shows that the definition evaluation function proposed in this paper meets the requirements of focusing accuracy,and that both sensitivity and noise resistance are better than traditional algorithms.(2)Regarding the division of focus windows,this subject proposes a window sampling method based on adaptive differential evolution algorithm.This method involve three steps:a)Combining the requirements of window sampling to set the initial population and control parameters of the evolutionary algorithm;b)Constructing the fitness function based on the sharpness distribution characteristics of the image edge,and c)realizing the search of target area based on the intelligent search algorithm.A large number of experiments have proved that the sampling method has higher positioning accuracy,while better meeting the needs of stability and applicability.(3)In order to improve the performance of auto-focusing search strategy,this paper proposes a new hybrid search strategy based on traditional methods.To be specific,first,the Fibonacci search method is applied to complete coarse filtering.Then,with the help of the adaptive mountain-climb searching algorithm,a precise positioning is achieved.These operations greatly improve the accuracy and efficiency of the search.In the end,a new auto-focusing method has been designed by combining the sharpness evaluation function and the selection strategy about focusing window proposed before.