VEHICLE INTERIOR CLASSIFICATION SYSTEM AND METHOD

摘要

A vehicle interior classification system and method in accordance with the present invention determines a classification relating to the interior of the vehicle, such as the occupancy status of a vehicle seat or the state of alertness of a vehicle driver, from one or more images of an appropriate portion of the interior of the vehicle acquired with an image capture device. The acquired images are preferably processed to limit the dynamic range of the images to obtain a resultant image which can comprise one or more regions of interest which are less than the total field of view of the image capture device. The resultant images are processed to extract information about features in the image and, in one embodiment, this processing is achieved with a two-dimensional complex discrete wavelet transform which produces a set of coefficients corresponding to the presence and/or location of the features in the resultant image. The set of coefficients produced with such a transform is potentially quite large and can be reduced, through described techniques, to a subset of the total number of coefficients, the members of the subset being selected for their ability to discriminate between the classifications defined for the system. By selecting a subset of the possible coefficients, computational requirements are reduced, as are hardware requirements in the system, such as memory. The selected set of coefficients (whether comprising all of the coefficients or a subset thereof) are provided to a classifier which processes the coefficients with a set of calibration vectors, that were determined when the system was calibrated, to determine the most probable classification for the portion of the vehicle interior.