Research Of Image Sharpness Assessment Algorithm For Autofocus

As imaging systems are widely used in various fields,such as: security monitoring,aerospace engineering,construction quality testing,medical image,and civilian infrastructure,it is crucial to capture a sharp image.The sharp images are captured only when scene is in sharp focus plane in general while defocused images will not provide valuable information and bring wrong information causing mis-calculation.Therefore,focusing is crucial for sharply imaging of imaging system.The early focusing mainly rely on manual.This method depend on personnel subjective judge to imaging quality,then determine whether is in focal plane.This focusing method has low efficiency and seriously restrict the development of imaging system.With extensive research on auto-focusing technology and image processing technology,auto-focusing technology arises at the historic moment based on image processing.Many researchers have proposed many methods,and some methods have been used in microscope and digital camera.The traditional sharpness assessment function are greatly influenced by image content and has poor applicability and stability,which performance is good only for a specific image.For large optical measurement equip-ment,due to its large field of view,widely focusing range,the richly observation target,auto-focusing is prone to failure with the traditional sharpness assessment algorithm.This paper mainly studies image sharpness assessment method for auto-focusing.By studying the auto-focusing technology,especially in development situation of auto-focusing technology based on image processing,the sharpness assessment method combined with fine focusing and coarse focus is put forward aiming at solving the shortcomings of the existing typical sharpness assessment method.This method adopt the method of rough focusing first making the focusing mechanism arrive at the focal plane,then using the precise focusing to find accurate position of the focus plane.The focusing method improve the efficiency of the focusing with the same focusing accuracy.At the same time,the method avoids that the focal status in depth is affected by noise into the local extreme value point.In order to improve the assessment performance of sharpness assessment algorithm,the essay focuses on the research of the principle of choosing the content of the window based on focusing window construction.In this paper,we study the sharpness assessment curve is influenced by three factors: the target proportion of the image,the information entropy,and the size of the interest area,then get the selection principles of interest area.In practice,according to the region of interest,choosing the part of focusing can improve the real-time performance,and the performance of the algorithm significantly.According to the characteristics of fine and coarse focusing,sharpness assessment algorithm is optimized.Three groups of different characteristics images are sharply assessed,and assessment results of various kinds of algorithms are analyzed.The optimal algorithms for different images,provide the basis for researchers selection to different sharpness assessment function of different application requirements.The sharpness assessment algorithm of the rough focusing selects the image quality assessment method for assessment.Firstly,images are spatial transformed to get the image normalization coefficient of means values contrast;then the statistical distribution is simulated by using the asymmetrical generalized Gaussian template to extract the feature information;and finally image sharpness is assessed using support vector machine(SVM)classification method.Though experimental verification,this method is correlated with subjective assessment method.To find the precise focal plane position,sharpness assessment of two or three frames images is only needed.This method of sharpness assessment greatly improve the efficiency of the focusing,and at the same time overcome the effects of image content on sharpness.