Home / Papers / Image processing apparatus, image processing method and image processing system

Image processing apparatus, image processing method and image processing system

88 Citations2011
薄叶亮子, 加藤典司, 熊泽幸夫
journal unavailable

An image processing apparatus acquires a captured image obtained by image-capturing a sample piece including a target cell having a nucleus, and determines whether or not the target cell is included in the rectangular area on the basis of whether orNot the image characteristic amount obtained from the established rectangular area satisfies an image characteristic amounts condition that has been machine-learned on the based of specimen images of positive and negative examples of thetarget cell.

Abstract

The tendency of being affected by variances of the samples of target cells, variances of image pickup conditions and the like is to be reduced. An image processing apparatus (10): acquires a captured image obtained by image-capturing a sample piece including a target cell having a nucleus; extracts, from the pixels included in the captured image, pixels, which correspond to a candidate of the nucleus, on the basis of a first condition that has been specified for at least one of a color and a brightness that the pixels corresponding to the candidate are to have; further extracts, from a group of linked pixels obtained by linking pixels adjacent to the extracted pixels, a group of linked pixels, which corresponds to a candidate of the target cell, on the basis of a second condition that has been specified for a magnitude and a shape that the group of linked pixels corresponding to the candidate is to have; establishes, in the captured image, a rectangular area having a given size and being centered at a pixel included in the extracted group of linked pixels; and determines whether or not the target cell is included in the rectangular area on the basis of whether or not the image characteristic amount obtained from the established rectangular area satisfies an image characteristic amount condition that has been machine-learned on the basis of specimen images of positive and negative examples of the target cell.