The image processing apparatus for detecting a moving object in a moving image includes a background generation unit configured to generate a background image of the moving image while updating the background image over time. The background generation unit includes a model derivation unit configured to derive a mixed distribution model having one or more distribution models for each pixel of interest, and a background value derivation unit configured to derive one or more background pixel values respectively corresponding to the one or more distribution models. The model derivation unit is configured to generate a new distribution model from pixel values of a plurality of pixels within a local region containing the pixel of interest in a first frame, and update the existing distribution model using a pixel value of the pixel of interest in a second frame that is different from the first frame.