A new method to detect object and estimate the position and the orientation from an image using a 3-D model having feature points

We propose an approach to detecting a three-dimensional (3-D) object and estimating its position and orientation. The proposed system can estimate the object's 3-D pose from a single input image by using 3-D model with feature points. The 3-D model is employed in order to represent a 3-D object, and the system can create 2-D models projected from various viewpoints. Then, the system compares each 2-D model with the input image and calculates the matching rate, taking notice of some partial features of the object. This object recognition can be regarded as an optimization problem: a maximum search problem of the model's matching rate. In the proposed system, a genetic algorithm (GA) is employed for this optimization. In the computer experiments, we took a car as an example of the object. We have tested various kinds of images and obtained excellent results.