Research On The Application Of Artificial Landmark Topology Pattern Construction And Recognition

Image-based positioning accuracy is usually affected by the image rotation and/or distortion due to the changes of the visual sensor position orientation.At the same time,image processing and positioning result is affected by the environment or other factors.So it's still a challenge to process the artificial landmark image and improve the accuracy and robustness of the positioning algorithm.In order to reduce the light influence,this paper proposed a new method of color image enhancement,in which the image was binary processed in RGB channels respectively and than merged together,the contrast between artificial landmark and background is obvious,which is in favor of the contour detection.And this paper proposed the elimination algorithm of broken contours and corners,which is used before property extraction and improve the positioning speed.According to the location relation among blocks in artificial landmark,this paper proposed artificial landmark topology pattern and the methods of construction and recognition.Its topology pattern is composed of the numbers of the block shapes,colors,shape and color of each block,and topological relation among the blocks,which has certain robustness for the artificial landmark image rotation and/or tilting.The positioning is based on the artificial landmark pattern recognition and the preset corresponding relation between artificial landmark and location.In this paper,the matching algorithm of artificial landmark topology pattern is proposed based on the pattern construction method,statistical pattern recognition and structural pattern recognition,which improve the accuracy and speed of positioning.In this paper,the artificial landmark is composed of 4 color blocks,and each block has six different shapes and four different colors,the experiment results show that the property of the color block were calculated accurately,the topology pattern of the artificial landmarks construction and recognition method is feasible and reliable.The accuracy,speed and precision of the visual positioning algorithm are higher,which can be used in monocular vision positioning.