To verify and evaluate results obtained by discriminating a plurality of areas of an image.
An image processing apparatus has a learning image reading unit 111 for acquiring image information of a learning image 20, a discriminator information determination unit 112 for determining discriminator information on the basis of the image information of the learning image 20 and a registered learning image kind 40, a learning image kind discrimination unit 113 for discriminating the kind of a pixel of the learning image 20 on the basis of the discriminator information, a target value setting unit 114 for setting a target value on the basis of a first difference quantity between the kind of the pixel of the learning image 20 and the registered learning image kind 40, a testing image reading unit 115 for acquiring image information of a testing image 30, a testing image kind discrimination unit 116 for discriminating the kind of a pixel of the testing image 30 on the basis of the image information of the testing image 30 and the discriminator information, and an evaluation unit 117 for evaluation whether a second difference quantity between the kind of the pixel of the testing image 30 and a registered testing image kind 50 is within an allowable range or not.
FUJITA YUSUKE
HASHIMOTO SHINICHI
TERAI SHUJI
SAKAIDA ISAO
Yoshiyuki Ishibashi