Abstract
Developing an efficient method of separating normally shaped somatic embryos from malformed ones is essential for automating the somatic embryo production procedures, and the method based on computer vision algorithm is expected as the strong candidate of the efficient method for this purpose.
Generalized Hough transform is known as the algorithm which is able to carry out very rapidly the procedure of template matching which is one of computer vision algorithm. BJcause the procedure of template matching seems to have several merits in application to the problem of selecting somatic embryos, e.g. the easiness of improvement of performance by changing templates, we tested the applicability of the procedure of template matching using the algorithm of generalized Hough transform in selecting somatic embryos by using real images of somatic embryos of carrot. As the result, our method could select out malformed somatic embryos fairly well, but it could not distinguish normally shaped somatic embryos and slightly malformed ones sufficiently.
