相关数值集算法在机器人图像识别中应用研究

RELATED VALUE SET ALGORITHM FOR ROBOT TO DISTINGUISH IMAGE

为能使计算机能自动从图像库中识别图像，产生了诸如ＱＢＩＣ［１］，结构分析［２，４］，图像区域分类法［５］等等许多算法．计算机从图像库中识别图像同现实中识别图像不同，机器人是移动的，图像大小随着移动大小发生变化，我们分析人识别图像的过程，得到图像主要特征的６个特点，从而得到相关数值集算法，根据这一算法，我们定义一集合代表图像主要特征，集合中每个元素用相关值表示，用这一算法从图像中捉取的特征对这６个特点来说是令人满意的．因此这一算法可以用做机器人图像识别中，主要特征可以用做图像知识表示．

In order to enable the computer to distinguish images automatically in an imagebase,many algorithms have been developed such as QBIC［1］,texture analysis［2,4］classification of image regions［5］,and so on．It i s different from a computer distinguishing image in an image-base for robot to distinguish images in real vorld.Robot is movable.The Size of the image that cap tured by the robot is varied with the distance between the robot's camera and the real picture.We have analyzed the process that human remembers and identifies an image have gott en six properties of the key feature in an imge.Then we have developed a related value set algorithm.With the algorithm,we defined set to represent the key feat ures of an image.The elements of the set are represented with the related value.The key features of an image The elements of the set are represented with the re lated value.The key features retrieved from an image with the algorithm can be u sed for robot to distinguish images.The key features can be used in image knowle dge representation.