Vehicle occupancy detection via single band infrared imaging

摘要

What is disclosed is a method for vehicle occupancy detection using a single band infrared imaging system. First, an infrared image of a vehicle intended to be processed for human occupancy detection is captured using a single band infrared camera set to a predefined wavelength band. A candidate sub-image is identified within the captured image. A cumulative histogram is formed using the reflectance values of each pixel in the candidate region. A threshold reflectance value is then determined from the cumulative histogram using a pre-defined cumulative occurrence fraction value which corresponds to a value equivalent to an average sized human face. Embodiments for setting the threshold reflectance value are disclosed. Thereafter, human occupants can be distinguished in the image from the vehicle's interior by comparing pixel reflectances in the sub-image against the threshold reflectance value.

主权项

1. A method for vehicle occupancy detection using a single band infrared imaging system operating at a pre-defined wavelength range of the electromagnetic spectrum, the method comprising:
capturing an infrared image of a vehicle intended to be processed for human occupancy detection, said image being captured with a single band infrared camera, said image comprising a plurality of pixels each having a reflectance value; identifying a candidate sub-image in said captured infrared image; dividing said candidate sub-image into at least two distinct sections such that one of said sections is likely to contain a driver of said vehicle; adjusting said cumulative occurrence fraction value to compensate for a reduced area of said section containing said driver relative to said candidate sub-image; forming a cumulative histogram using reflectance values of pixels in said section containing said driver; determining a threshold reflectance value from said cumulative histogram using said adjusted cumulative occurrence fraction value; differentiating at least one human occupant in said section of said vehicle containing said driver by comparing reflectance values of pixels in said section containing said driver against said threshold reflectance value; and differentiating human occupants, if any are present, from the interior of said vehicle, in the remaining sections, using said threshold reflectance value determined from said section containing said driver.