Application of image entropy evaluation function for the leveling of large aperture components in auto defects detecting

Abstract

In large aperture component’s dark-field scattering defects imaging system, the component’s size is large and part with a wedge. When the component is in the completely level position, the surface defects image can be clearly acquired by a high magnification microscope. Otherwise, fuzzy defects image would be gained because of defocusing which makes digital identification can’t be able to be done. For the problem of leveling large aperture, wedge component, this paper proposes a method that using image information entropy as focusing evaluation function for leveling large aperture components. Firstly, in three different points of component surface acquiring multi-images by the same continuous steps. Then calculating the images’ entropy and fitting a curve to it. Based on minimum image information entropy value criterion, the focal plane can be found and each point’s defocusingamount of the fist acquisition position can be gained. Relay on the relation model of acquisition points, adjust points and defocusingamount that has been built, each adjust point’s adjustment can be got. The component’s level position can be achieved by adjusting the adjust points. In the experiment that using a high magnification (of 16) microscope scans over the whole surface of the component with the size of 430mm×430mm. The image microscope is always in the depth of focus which shows that the leveling precision has achieved 20μm. Until now, this method has been successfully used in large aperture component’s dark-field scattering defects imaging system.