Extracting a focused image from several out of focus micromechanical structure images

An all-focused image of a micromechanical structure can be extracted from several microscopic images with different focal planes. Spreading of the out-of-focus regions over the background, which shows the out-of-focus object bigger than its actual size, causes errors when a sharpness function is applied. The focused area with low contrast is also one of the difficulties faced in order to easily distinguish them from the out-of-focus regions. We propose a method to estimate the focused section of each image and create a final focused image by concatenating these areas. We let the size of the tested areas vary in order to obtain a more accurate result. The method finds focused pixels among the images using a voting method and pastes them to the final image. This approach does not rely on any information of the microscope. No a priori knowledge of an optical system is needed.