An image processing apparatus includes an image clipping unit, a feature extracting unit, a candidate identifying unit, and a detecting unit. The image clipping unit clips a window image from a predetermined position of an original image. The feature extracting unit extracts a feature value of the window image on the basis of a predetermined criterion. The candidate identifying unit determines, on the basis of the feature value, whether the window image satisfies a predetermined condition for a candidate including a detection target. The detecting unit determines whether the window image includes the detection target if the window image satisfies the predetermined condition.