| 摘要 |
PROBLEM TO BE SOLVED: To reduce data amount and processing time without lowering identification performance when an object is identified from an image.SOLUTION: A first feature quantity extraction section 24 extracts a first feature quantity vector H having a value associated with physical quantity obtained by dividing feature quantity into a plurality of feature quantity as an element from a window image to be a processing object. A criterion determination section 26 determines a value larger than an average of each element of the first feature quantity vector H as a criterion value th. A second feature quantity extraction section 28 determines whether or not a feature quantity value hof each element of the first feature quantity vector H is the criterion value th or more and extracts a second feature quantity vector X setting the feature quantity value hof the element to be h<th as 0. An object identification section 30 identifies whether or not a window image is an image showing an identification object on the basis of the second feature quantity vector X and an identification model stored in an identification model storage section 60. |
