3D Microscopic Measurement of Seedlings Using a Shape-from-Focus Method

Abstract

The shape-from-focus algorithm, a passive method, was modified for microscopic measurement of 3 D shape and texture mapping of intact petunia seedlings with coarse surface. First, both max-min and least-squares (linear regression) operators newly developed to estimate the focus measure index were compared with the sum-modified-Laplacian operator proposed by Nayar and Nakagawa. In the new operators a deformed mask was used for shortening of calculation time. Also, RGB full color images as well as gray image were used as a series of original 2-D images measured at consecutive focused planes by a computerized light microscope. As a result, the focused range image, i.e. 3-D shape estimated using the least-squares operator and RGB images was the best of all. This combination was effective for estimating the focused range in the case of not only the coarse texture but also the texture with large gradient in gray level such as the edge. Secondly, a focused color image was interpolated from a series of original images using the range image. Also, wire frame and texture mapping images were made using the range and focused color images. As a result, it was possible to make a clear color image of the seedling. Furthermore, it made easy to observe 3D color images of the seedling from any direction by operating a mouse.