“UD-ETR Based Restoration & CNN Approach for Underwater Object Detection from Multimedia Data”

In this paper, the work has been done in the field of underwater world, which includes restoration of the obtained images and then the detection of objects. The identification of the objects is performed in order to check the presence of items in the video and to unequivocally find a particular object. This is the area of interest for researchers because of the presence of various species under water. The images which are taken in the water lacks some factors such as: non-uniform lighting, low contrast, turbid water etc. which leads to the generation of improper outputs. These issues give rise to the involvement of restoration technique, which is a fundamental technique used for removing some severe abnormalities from the obtained images such as: low contrast, blurring etc. The proposed algorithm UD-ETR (UDCP based Energy Transmission Restoration) is used to restore the green channel images. It contains multiple stages of restoration which ultimately gives final restored image. Further, CNN (Convolutional Neural Network) is used for detecting the fish from the restored image. This algorithm is implemented using the MATLAB tool and the efficiency obtained shows its capability.