Related value set algorithm for robot to distinguish image

The image recognition from image database is different from robot vision in real world because the robot concerned is movable. The size of the image that captured by the robot varies with the distance between robot´s camera and the real picture, or object. We analyzed the process of how human remembers and identifies an image. We derived eight properties of the key feature in an image. We then developed a related value set algorithm. With the algorithm, we defined a set to represent the key features of an image. The elements of the set are represented with the related value. The key features retrieved from an image with the algorithm satisfied the eight properties, thus the algorithm can be used for robot to distinguish images or object. The key features can also be used in image knowledge representation