An image processing device includes: a predicting unit configured to, from a position of an object in a previous frame with respect to a current frame, predict a position of the object in the current frame and identify a prediction area; a first matching unit configured to perform template matching in the prediction area in the current frame using a first template regarding the object in the previous frame, to detect the object, when the object is present in a first distance area; and a second matching unit configured to perform template matching in the prediction area in the current frame using a second template which is different than the first template and which is regarding the object in the previous frame, to detect the object, when the object is present in a second distance area which is farther than the first distance area.