In this paper, a new method of marine object recognition using blob analysis has been proposed, which is suitable to general objects recognition. A powerful foreground blob analysis is proposed to classify frontal areas. Conventionally, the main focus of the objects is recognized by prepared researchers through towed nets and human perception, which make much cost and hazard administrators and animals. Specific marine objects, box jellyfish and ocean snake, are effectively recognized in this work. Experiments conducted on picture datasets gathered by the Australian Institute of Marine Science (AIMS) demonstrate the adequacy of the proposed strategy.