| 摘要 |
PROBLEM TO BE SOLVED: To set an optimal value for each of a learning image processing parameter and an input image processing parameter to be used for image recognition processing for detecting the image of a target object from an input image.SOLUTION: A learning part 1110 generates a dictionary for image detection of a target object from an image obtained by performing image processing based on a learning image processing parameter to a learning image. A detection part 1090 detects a target object on the basis of the dictionary from an image obtained by performing image processing based on an input image processing parameter to an input image. An evaluation part 1080 compares a model image generated from the detection result with the edge part of the input image, and calculates the evaluation value of the detection result. A learning image processing parameter setting part 1070 sets a learning image processing parameter such that the evaluation value becomes the highest (1070). Also, an input image processing parameter setting part 1040 sets the input image processing parameter such that the evaluation value corresponding to the detection result using the dictionary generated on the basis of the set learning image processing parameter becomes the highest. |
