The Application Study On Image Recognization Technology In Hand-claw Perception System Of Underwater Robot

Following extensive marine development, the underwater robot is used widely worldwide. The operating system of the underwater robot can determine the globality of the underwater robot, but the manipulator under water does not dispose too many sensors basically at present, the information not reaching the hand claw perceives competence. To realize underwater robot without cable (autonomy robot ) that is engaged in more complicated unknown environment to operate acting on one's own must break through key difficult problem of hand claw information while perceiving . The application in the underwater robot of recognition technology of picture is the important content of realizing the perceiving system of claw information of hand. This paper regards realizing the environmental goal localization based on recognition technology of picture as the goal ,to special circumstances under water , designs and succeeds in developing one hand claw information perceiving system of the underwater robot based on CCD vision sensing under water. This system is to make up by vision sensing , pretreatment of the picture and PUMA560 manipulator , hand claw under water. According to function of system require , testing software of hand claw information perceiving system of the underwater robot is written by utilizing Visual C++ 6.0 under the environment of Windows, in order to realizing the discern orientating demand of environmental goal under water. While shooting the picture of environmental goal under water by adopting CCD lens, because of goal surface under water , auxiliary light source , interference of law of perturbation of environment under water, and the definition of the goal under water in the picture of the hand claw moving that bubble influence produced absorbs, so difficult to get the rash picture of goal under more water than the high one of letter. At first it must strain picture wave and make an uproar , edge measuring or two value ,etc. Make environmental goal picture characteristic outstanding , and could by hand claw information perceiving system go on recognition and location of environmental goal under water. This paper has adopted all kinds of picture pretreatment methods to deal with it through analyzing the environmental goal picture under water gathered in real time. Through the comparison of these methods, the selection hitting value strains the wave law and deals with noise reduction, have been adopted many yardstick of small wave to vary and draw on the environmental goal edge under water since meeting two value treatment of law of threshold value, the result has keep most details, has got the pretreatment picture that is helpful to the environmental goal characteristic under water. On the foundation of analyzing hand claw information perceiving system of the underwater robot based on CCD vision sensing in theory, adopting PUMA560 as drive platform of hand claw information perceiving system under water, has improved the systematic precision greatly ,and adopting the "dark case" thought of the coordinate to shine upon picture coordinate directly, without standard internal and external parameter of CCD under water, realizes discernment and location of the goal under water, it reduced the complexity of the experiment .Indicate through the hand-eye vision goal Localization experiment: 1) error of the system on X-Y level <+1mm, the error of orientating in Z axle direction <+1. 5mm; 2)The system is orientated the error in the repetition with the direction of the axle <0. 2mm, the repetition in the direction of the axle is orientated the error <0. 8mm. Considered the repetition of PUMA560 manipulator orients the error <+0. 3mm, these reached higher precision after proving the system, meeting the demand of the information perceiving system of the hand claw...