The present disclosure pertains to an image processing device capable of continuously tracking an object in a video, an image processing method, and a program. An image processing device according to one embodiment of the present disclosure is equipped with: a plurality of object characteristic calculation units for calculating the object characteristics of an object in a video from the current frame of the video, on the basis of standards which differ by unit; a characteristic selection unit for selecting an object characteristic to be used from among the plurality of object characteristics calculated on the basis of the different standards, by using object attribute information pertaining to the object in a previous frame of the video; an object information calculation unit for calculating object information pertaining to the object in the current frame, by using the selected characteristic and the object information pertaining to the object in the previous frame; and an object attribute information calculation unit for calculating the object attribute information pertaining to the object in the current frame, on the basis of the calculated object information pertaining to the object in the current frame. This disclosure is applicable to surveillance systems.