Image enhancement and compression of deep-sea floor image for acoustic transmission

摘要

One of the important missions of AUVs is to take deep-sea images and make maps of sea floor to contribute the understanding of deep-sea biology and geology. The sea floor images include lots of information such as distributions of creatures, minerals and so on. The next mission expected for AUVs is to bring the samples back concurrently with taking interesting photos, autonomously. However, this mission is difficult to perform fully-autonomous without the knowledge of biologists or geologists, as the AUVs don't know which objects are interesting or important for them and it takes long time to teach AUVs all features of target samples. It is necessary for AUVs to work semi-autonomously by collaborating with scientists on the ship and getting commands to indicate the important samples. In this research, we proposed an image enhancement and compression method for acoustic transmission with limited communication density. The proposed method enhances the effect of light attenuation in the deep-sea floor image and reduces the image depth from 24-bit of original images to 4-bit. The method is evaluated by transmission experiment using a new AUV "TUNA-SAND 2".