Research Of HD IP Camera Autofocus Technology

In recent years, the rapid development of digitization and networking, as well assignificantly enhanced safety awareness of the people, has made a major breakthroughin the development in the field of security. HD network camera (HD IP camera) hasthe advantages of camera higher clarity, better interactivity, and more intelligent thanprevious analog cameras. Together with its lower and lower price, HD IP camera haswidely applied to various fields. Application of HD IP camera in high-end occasionshas occupied the main position, including: highway, Banks, telemedicine, securitybarrier, and so on. It caused great favor and attention of big security manufacturersand users. In a very short period of time, HD IP camera has become one of the leadersin the field of security.Automatic focusing is a very important technology of HD IP camera intelligentapplication. Autofocus technology directly affects the quality of the obtained images.For this reason, automatic focusing has become a key technology to research for along time. With the rapid development of technology, the improvement of imagingequipment resolution and the popularity of intelligent application, the latesttechnology is gradually applied to high definition webcam. These advancedtechniques urgently need the support of auto focus technology. It has promoted therapid development of the automatic focusing technology. Its application hastranscended the traditional digital camera, components inspection, medical imaging,space exploration, microscopic imaging, and extends to security, remote medicaltreatment, machine vision, etc. For automatic focusing, which contains two maintechnologies: one is focused on the selection of the evaluation function, the other is apeak search algorithm implementation.Automatic focusing contains two main technologies: one is the selection of theevaluation function, the other is a peak search algorithm implementation.In this paper, we study the focusing evaluation function includes: gray variancefunction, gradient function, frequency domain function, entropy function,autocorrelation function, and AF engines. In view of the big computation on thetraditional focusing evaluation function, we choose TI Company DM368processorcomes with the hardware implementation of IIR filter as focusing evaluation function.Through comparing and analyzing all kinds of focusing evaluation function by analytic hierarchy process (AHP) show that the AF engine focusing evaluationfunction is better meet the application of HD IP camera application. In addition, theconfigurable features of AF engine can effectively solve the problems in intelligentapplication. And the configurable features make it easier to combine automaticfocusing algorithm with intelligent algorithms such as scene recognition, facedetection and so on. It meet the development trend of intelligent equipment.Peak searching algorithm is improved based on the analysis of the traditionalsearch algorithm and focusing curve characteristics. It named the focusing area whichis near the best imaging plane focused area, and named the focusing area which is farfrom the best imaging plane defocused area. Through the analysis, we know that thefocus curve in defocused area and focus area present their features respectively. In thefocus area, we choose traditional monotonous judgment method as automatic focusingmethod. In the defocus area, we choose peak searching algorithm which is based onGM (1,1) prediction model as automatic focusing method. The proposed algorithmcan effectively solve the problems such as video quality changed repeatedly, longpeak searching time and defocusing area local peaked interference problem. Inaddition, this paper also introduces the other associated with AF technology, includingreturn to check the maximum method and the variable step search strategy and theanalysis of the scene changes, etc.Back to check the maximum value method can effectively solve the problemwhich is caused by the scene changes in the AF process. Variable step searchingstrategy use real-time step changing strategy to improve the average speed in theprocess of peak searching. And this method can ensure the real-time performance ofAF. Changed scenario analysis can effectively trigger autofocus again.This article choose TI company TMS320DM368processor as the coreexperiment board, using its own hardware implementation of two-way IIR filter asfocusing evaluation function. Peak searching algorithm implementation choices in PCoperation, sent via serial interface is focused on the command, finally complete theautomatic focusing.This article choose the TMS320DM368processor which is produced by TICompany as the core experiment board, using its two-way IIR filter which ishardware implementation as focusing evaluation function. Peak searching algorithm isrun in a PC, and sent AF command via serial interface, finally complete the automaticfocusing. Experiments show that the AF engine and monotonous judgment method andGM (1,1) forecasting model with the combination of traditional peak searchingalgorithm is focusing evaluation function and search algorithm with the combinationof focusing algorithm has better real-time performance and accuracy, and is suitablefor automatic focusing of HD IP camera.Experiments show that the algorithm which use AF engine as the focusingevaluation function and use monotonous judgment method and GM (1,1) forecastingmodel as peak searching algorithm is better than the algorithm which use traditionalfocusing evaluation function and peak searching algorithm. And it has better real-timeperformance and accuracy, and is suitable for automatic focusing of HD IP camera.