Ultra-fast parallel computation method for integrated image based image segmentation and its media recording thereof

摘要

The present invention generates an integral image from a given ultra-high-resolution graylevel image and uses it to perform image segmentation using the value of a Difference of Averages (DoA) filter. It relates to a parallel computation method that generates (trianry image) at high speed. An ultra-high-speed parallel operation method for image segmentation using an integral image according to the present invention includes: a padding image from a given input image and a simultaneous generating unit of an integral image thereof; a mean difference filter unit using the calculated value of the integral image; and an image segmentation ternary image generation unit using the values generated from the mean difference filter unit; includes In the case of implementing the integral image-based image segmentation method as software in a computer equipped with a multi-core CPU as an embodiment according to the present invention, a multi-GPU that requires additional cost because it can be implemented in real time only with a multi-core CPU There is an effect that I do not need to use a special embedded (embedded). In addition, there is an effect that can be used for image quality assessment to determine whether there is an abnormality in the camera and lighting, which are essential components of a machine vision inspection system, using the integral image.