Study on Approach of 3-D Object's Visual Recognition

Summarizing an approach basing on 2-D image to recognize 3-D objects. Extracting the object's valid area by way of complying with color characteristics,using a chain code to describe the shape characteristics of 3-D objects in the area,and the object's shape characteristic vector is formed.The vector characteristic is used for object-matching recognizing of what is designed for input of BP neural network.The network output is a kind of subordination of certain objects.Effectiveness of the method is proved through an experiment of 2-D color images of different 3-D objects in an environment under various luminous conditions.