Information processing device for detection of background of image
A control unit executes: acquiring an actual image that is an image captured by the imaging unit; acquiring a calibration image that is an image captured by the imaging unit in the past and serving as a reference for the current background; generating a model for estimating a region where a shadow exists and correcting the color tone; and adding a shadow corresponding to the actual image to the calibration image based on the calibration image and the model and correcting the color tone and generating an estimated background image, which is a corrected image and is a background image corresponding to the actual image.
1 . An information processing device comprising a control unit configured to:
acquire an actual image that is an image captured by an imaging unit;
acquire a calibration image that is an image previously captured by the imaging unit and that serves as a reference for a present background;
generate a model for estimating a region in which a shadow is present and performing color tone correction based on the actual image;
generate an estimated background image based on the calibration image and the model, the estimated background image being an image obtained by adding a shadow corresponding to the actual image to the calibration image and performing color tone correction on the calibration image and being a background image corresponding to the actual image;
acquire an ideal background image that is an image in an initial state;
generate a background replacement image based on the actual image, the estimated background image, and the ideal background image, the background replacement image being an image obtained by replacing a background portion of the actual image with a corresponding portion of the ideal background image; and
determine presence or absence of foreign matter based on the background replacement image.
2 . The information processing device according to claim 1 , further comprising
a storage unit that stores a trained machine learning model for determining the presence or absence of the foreign matter based on the ideal background image, wherein
the control unit is configured to determine the presence or absence of the foreign matter by inputting the background replacement image into the trained machine learning model.
3 . The information processing device according to claim 1 , wherein the imaging unit is arranged inside a vehicle, and captures an image of at least an entrance/exit of the vehicle.
4 . The information processing device according to claim 1 , wherein the control unit is configured to generate the model by:
extracting a shadow candidate region, as a candidate for a shadow region, from a brightness distribution of the actual image;
extracting an estimated shadow region from the shadow candidate region through a comparison with an estimated background image generated previously; and
determining a correction parameter for approximating a color tone of the calibration image to a color tone of the actual image through a comparison between the actual image and the calibration image.