LEARNING DEVICE, LEARNING METHOD, AND LEARNING PROGRAM

摘要

A motion setting section (61) sets a motion amount and a motion direction for determining a processing coefficient. A student image creating section (62) creates a student image by adding motion blur to a teacher image after changing at least one of the motion amount and the motion direction at a specific ratio not only by using the set motion amount and the set motion direction and creates a student image by not adding motion blur. A predicting tap extracting section (64) extracts at least the pixel values in the student image at almost the same position as the space position of the pixel of interest in the teacher image so as to extract the main term mainly including the component of the pixel of interest. A processing coefficient generating section (65) generates a processing coefficient for predicting the pixel of interest in the teacher image on the basis of the pixel values of the extracted pixels according to the relation between the extracted pixel and the pixel of interest in the teacher image. A processing coefficient for robust motion blur removal for shift of the motion vector can be generated by learning.