主权项 |
1. An image processing method for use by an image processing system including a processor executing a program stored in a memory and a storage unit for holding a plurality of images, the image processing method comprising:
an acquisition step of acquiring, by the processor, data containing an image; a first extraction step of extracting, by the processor, a first image region from the acquired data in accordance with a type of software used for generating the acquired data; a second extraction step of extracting, by the processor, a second image region that is the same as or similar to each of the plurality of images held in the storage unit from the acquired data, wherein the image similarity is based upon comparing the acquired data and each of the plurality of images held in the storage unit to calculate a characteristic of compared image features which is compared to a predetermined threshold; a third extraction step of extracting, by the processor, a third image region that is the same as or similar to each of the plurality of images held in the storage unit from the acquired data by comparing an image feature amount of the acquired data and an image feature amount of each of the plurality of images held in the storage unit in accordance with the characteristic of the compared image features relative to a predetermined threshold; and a step of identifying, by the processor, an image in the acquired data to be stored in the storage unit based on the first image region, the second image region and the third image region, and reliabilities set to the first extraction step, the second extraction step and the third extraction step in accordance with another predetermined threshold, wherein the step of identifying the image in the acquired data to be stored in the storage unit comprises: calculating, by the processor, a ratio representing mutual overlaps among the first image region, the second image region and the third image region; and identifying, by the processor, the image to be stored in the storage unit out of the first image region, the second image region, and the third image region based on the calculated ratio representing the mutual overlaps, and the reliabilities set to the first extraction step, the second extraction step and the third extraction step. |