Occupant Pose and Location Detect for Intelligent Airbag System Based on Computer Vision

Airbag deployments during automobile accidents can cause injury and even death if the occupant is an improper size or in an improper position. Because of this, steps have been taken to make airbag deployment safer by adding adaptive deployment decision capabilities. This paper presents a new method to obtain occupants´ pose and location by using a CCD camera fixed in a cabin. The face location is the target of our initial focus, as it provides a good estimate of the occupant´s head position and is the most critical part of the body in airbag deployment. The occupants´ head locations are detected through obtaining their face position. First, establish criterion templates to represent different size of occupants, that is, children, the 5^th percentage of female, the 50^th percentage of male and the 90^th percentage of male. Then compare the occupants´ stature with the criterion templates. The airbag can be fully deployed, partially deployed or nor deployed at all according to occupants´ size and location. This arithmetic has the advantage of simple, accuracy, real time and less sensitive to illumination. Experimental results show that the recognition accuracy of the proposed approach is better. The total correct classification rate is 96.5%. It has great significance to maximum protect occupants´ lives according to occupants´ stature to control the deployment of airbag.