Integration of enhanced background filtering and wavelet fusion for high visibility and detection rate of deep sea underwater image of underwater vehicle

This paper presents an enhanced technique for contrast and visibility improvement for deep sea underwater image which is normally used for underwater robot. The proposed technique uses an integration approach of enhanced background filtering and wavelet fusion methods (EBFWF). The novelty lies in this case in its methodology and capability of the proposed approach to minimize negative underwater effects such as blue and green color casts, low contrast, and low visibility in comparison with other state-of-the-art methods. The proposed method consists of a few steps that aims to eliminate negative effects and thus improving the contrast and visibility of underwater image. This purpose is carried out to provide a better platform for object detection and recognition processes. The input image is first sharpen before the low frequency background is removed. This minimizes the probability of image data to be regarded as noise in the consequences processes' steps. Image histograms are then mapped based on the intermediate color channel to reduce the gap between the inferior and dominant color channels. Wavelet fusion is applied followed by adaptive local histogram specification process. Based on the conduced tests, the proposed EBFWF technique, computationally, more effective and significant in improving the overall underwater image quality. The resultant images processed through the proposed approach could be further used for detection and recognition to extract more valuable information.