Research And Implementation Of Digital Auto-focusing System Based On ARM-linux

Nowadays due to the rapid development of electronic technology and optical technology,photographic and monitoring equipments are highly required in the field of both hardware andsoftware, which makes these equipments becoming the combination of the most advancedtechnologies around the world. Auto-focus, as a key factor of monitoring equipments, hasearned a high degree of attention in this walk of life. Many enterprises devote huge scientificresearch capability in it with the purpose of providing much more perfect and efficientauto-focus technology for customers, thus to obtain more economic benefits from this.The aim of this dissertation is to study, bring forward and achieve an ARM-Linux baseand superior auto-focus technology.The dissertation first studies four important parameters which are focal length of the lens"", relative aperture, field of view "2ω" and resolution. Then it analyzes the moving processof a zoom lens shot under lens groups in the processing of zoom. Lots of test about how toevaluate the quality of digital images are being carried out when lights go through CCD viacamera shot. Based on these tests, three functions as Sobel operator, energy gradient operatorand image spectrum amplitude of weighted normalization are chosen as evaluation functionsunder different stages.Trilogy of autofocusing algorithm is being presented in this dissertation as “trilogy1”“trilogy2” and “trilogy3”. Trilogy1deals with issues about narrowing focus range and movingto the near position of quasi-focus in large steps, while trilogy2concerns of issues aboutfinding focus through small steps, which leads the whole system to the quasi-focus position,and trilogy3solves issues about when to start focusing based on the dynamic monitoring onthe surrounding circumstances with the prerequisite of quasi-focus position. Also, these threemodules can transform into each other under different conditions. In the mean time,auto-focus, auto-exposure and auto-white balance, known as “3As”, are of mutual restraint,thus,3As will be overall planned, figured out and solved together.There are three innovations on the mentioned trilogy of autofocusing algorithm in thisdissertation as following:1) When the camera lens is of higher magnification, system will evaluate the newimage range which is formed through extracted and interlaced lines and rows. This way canovercome the shortage of edge information of certain images, and will improve the validityand efficiency of autofocusing. 2) Different functions will serve as evaluation function under different stages, whichwill lead to better focusing results. For example, as to trilogy1and trilogy2, energy gradientoperator will be used when light intensity is high; otherwise Sobel operator will be used.When trilogy3is used, image spectrum amplitude of weighted normalization will be used asevaluation function.3) Trilogy of autofocusing algorithm itself is one of the innovations of this dissertation.It analyzes3different statuses, the difference between them, and how to apply to differentsolutions according to different conditions.In the end, analysis on the results will be made based on four standards as speed,effectiveness, stability and comfort.