Study On The Auto-focusing Technology Of Digital Image And Realization For An Auto-focusing System

From the middle period of the last century to now, the applications of photoelectricity imaging systems have been extended, which were limited to perform the function of recording static images at the early time and now come into use to analyze a target object. And nowadays photoelectricity imaging systems are applied to many fields, such as industry, agriculture, medicine, military affairs and laboratory research etc. In a photoelectricity imaging system, precisely focusing is a primary problem to be solved, which greatly affects the image quality in back end and the efficiency of coming performance with image. There is a close relationship between the development of the auto-focusing technology and the development of cameras. And the development of the auto-focusing technology is represented by the innovations in the design of auto-focusing mechanism inside a camera, which embodies the level of integration of optics, mechanics and electronics, and directly affects the performance of a camera.Auto-focus is an important function in all kinds of imaging systems. It has widely application in many fields. In our country, the research and development about auto-focus is poor, but we are facing kinds of application requirement. The image processing is a rapidly developing technology. It promotes the development of the auto-focus technology greatly. And numeralization and intelligentization has being a tendence of the auto-focus technology. For these conditions, providing the auto-focus means based on image processing technology and developing the auto-focusing systems with excellent performance will have important significance. It not only can push the advancement of the integrated technology of photoelectricity and mechanics, develop the industry of contemporary optical instruments, but also can progress the industry of digital imaging systems if the auto-focusing modules are integrated into the imaging systems. Finally it makes our imaging systems to have stronger competition in domestic and international markets. The research subject comes from the Zhejiang provincial grand item of science & technology "The auto-focusing image system based digital technology" (No. 021105778, KYZ011103002) and the national nature science foundation item of China "Study on the auto-focusing method and system based on neural network identification and control" (No. 60672063).The auto-focusing method based on image technology is different from the traditional auto-focusing method entirely. The traditional ones must depend on some special assistant units. They use these units to measure distance or find focus and realize auto-focus. However, the auto-focus based on image technology is completed by analyzing the image quality directly, judging the imaging state from the current image quality, and adjusting correctly the focusing distance supervised by the imaging state. The auto-focusing device is an important part in a photoelectricity imaging system. The paper dissertates all kinds of auto-focusing methods considering the process and actuality of the auto-focusing technology, and then analyzes and compares their characteristics. It comes that the auto-focusing method based on image technology has developmental potential. Then the paper describes a developing trend of the auto-focusing technology and thinks that high integration, intelligence, low power consumption and high processing speed are the coming remarkable character of the auto-focusing technology.From the essential principle, the auto-focus can be divided into two classes. The one is the method based on measuring the distance between lens and an object. The other is the method based on measuring the focusing condition on a focusing screen. In the recent years, the auto-focusing method based on image treatment appeared. The paper dissertates three kings of auto-focusing methods and emphatically discusses the essential principle and the compositive modules of auto-focusing systems and the due property of focusing evaluation functions.The preprocess technology of digital image is a basis and prerequisite of the auto-focusing method based on image treatment. The paper discusses the image numeralization and its character, presents the two greyness-emendation technology based on the linear transform and the histogram equilibrium of image greyness-value considering the ray sensitivity of CMOS image sensors, introduces the different realization methods against positive noise and multiple noise. As to positive noise, the mean filter of neighborhoods and median filter can be used. According to the result of the experiments, the former will weaken the image edge and will not be good for reducing noise of spiced salts. The latter has a good effect to reduce positive noise and to conserve the image edge. As for the reduction of multiple noise, the homomorphic filter can be chosen, which separates signal and noise, filters noise, and then restores original signal. The filter's effect is determined by its design.The image quality analysis and evaluation is the key technology. In the thesis, three image quality analysis methods are presented detailedly.(1) The first one is the image quality evaluation based on image contrast. It evaluates the image definition at temporal field, frequency field and information entropy. By comparing their performance in the number of calculation and the time consumption of calculation, we find out the two better functions that are respectively the Brenner function and the absolute variance function. On the foundation, the paper presents the two improved algorithms based on the greyness-contrast variation and the autocorrelation function. The simulation results indicate that the two methods are simple and practical, not only have the merits of special evaluation functions but also overcome the demerits of strong sensitivity to noise and high demand for focusing area of special evaluation functions.(2) The next one is the image quality evaluation based on image power spectrum. It supposes nature scenes have the same power spectrum, introduces a human vision system and includes a Winner de-noise filter. It evaluates a picture and gets an IQM (Image Quality Measure) value. The IQM value is highly relative to the estimation result through human vision system. The IQM method can evaluate image quality directly. The log(IQM) value and the human vision evaluation can satisfy an approximate linear relation. So its evaluation results can represent the human vision evaluation and be used for the references of other applications and the direct focusing control. The IQM value can control the adjustive focusing distance definitely in focusing applications. It decreases searching process compared with the image contrast method and has faster response when the demand for precision is not strict.(3) The last one is the image quality evaluation based on wavelet analysis and neural network. It analyzes image from different frequency resolutions through wavelet theory, and abstracts the image characteristics from the detail information through statistics analysis, then uses an artificial neural network for quality pattern recognition in order to get image quality grades. The method is based on the biology vision mechanism. On the foundation analyzing the focusing relation of human's eyes and imaging systems, it includes human subjective factors into focusing evaluation rules to improve the focusing effect of imaging systems by the nonlinear character of neural network. The experiment results indicate that the recognition rate of this method is good.An auto-focusing system carries on partitioning and selecting of image in a focusing process to reduce the data quantity to be process, so the calculation decreases greatly. The paper introduces detailedly the focusing process of the complicated image contrast method, presents the method comparing thresholds to overcome disturbing influence in the course of focusing search by analyzing and summarizing the exceptional phenomena which may happen in the imaging process, and shows briefly the focusing courses of the power spectrum method and the wavelet analysis & neural network method. The focusing realization for the contrast method is an approaching course of repetitious regulations, so it can reach to a high precision. But it takes time because of calculating over again after every regulation. As for the power spectrum method or the wavelet analysis & neural network method, they move a special distance relative to the image quality level directly. They have more rapid response although their precision is a little lower.Depending on the above research of the auto-focusing technology, the paper presents the realization scheme of an auto-focusing system based on FPGA (Field Programmable Gate Array), analyzes the characteristics and demands for the optics imaging module, the image collection module, the image processing module, the control module and the driving module, designs and confirms the concrete realization scheme and systemic parts, and studies some key problems of the system realization. Those problems include the time-delay character of all modules, the realization of FFT algorithms, the interface realization of I~2C, the chip configuration of image collection, the interface realization of VGA and the chip configuration of FPGA. Finally, the paper analyzes detailedly the problems and reasons of the system realization, presents the improved system structure, and designates the direction of improving systems.