Robots working in an underwater dynamic environ-ment need complex visual perception to adapt to the environment, which requires emergency planning ability to deal with various visual pollution emergencies and improve the reliability of oper-ation. An emergency strategy is proposed for the visual pollution of the operating equipment, which used case-based reasoning (CBR). The correct detection and extraction of pollutant char-acteristic information are realized by similarity matching. Some general visual contaminated cases are used to build the case library of visual contamination. The image de noising strategy of case-based reasoning and the cleaning strategy of bionic cleaning equipment are analyzed and implemented. One bionic visual decontamination equipment is designed to clean the pollutants. Based on CBR for different pollution situations, the feasibility of several schemes is verified.