A Contrast-Guided Approach for the Enhancement of Low-Lighting Underwater Images

Abstract

Underwater images are often acquired in sub-optimal lighting conditions, in particular at profound depths where the absence of natural light demands the use of artificial lighting. Low-lighting images impose a challenge for both manual and automated analysis, since regions of interest can have low visibility. A new framework capable of significantly enhancing these images is proposed in this article. The framework is based on a novel dehazing mechanism that considers local contrast information in the input images, and offers a solution to three common disadvantages of current single image dehazing methods: oversaturation of radiance, lack of scale-invariance and creation of halos. A novel low-lighting underwater image dataset, OceanDark, is introduced to assist in the development and evaluation of the proposed framework. Experimental results and a comparison with other underwater-specific image enhancement methods show that the proposed framework can be used for significantly improving the visibility in low-lighting underwater images of different scales, without creating undesired dehazing artifacts.

L2UWE: A Framework for the Efficient Enhancement of Low-Light Underwater Images Using Local Contrast and Multi-Scale Fusion

“…Sub-section 2.1 discusses works in three areas of relevance to the development of L 2 UWE: underwater image enhancement, aerial image dehazing and low-light image enhancement. The sub-section that follows, 2.2, summarizes the single image dehazing framework of He et al [24], the multi-scale fusion approach of Ancuti and Ancuti [7], and the contrast-guided low-light underwater enhancement system of Marques et al [39].…”

“…Cabled ocean observatories have captured thousands of hours of underwater videos from fixed and mobile platforms [39]. The manual interpretation of these data requires a prohibitive amount of time, highlighting the necessity for semiand fully-automated methods for the annotation of marine imagery.…”

“…Its novelty is based on the observation that low-light scenes present particular local illumination profiles that can be efficiently inferred by considering local levels of contrast. We propose two contrast-guided atmospheric illumination models that can harvest the advantages of underwater darkness removal systems such as [39], while preserving colors and enhancing important visual features (e.g., edges, textures). By combining these outputs via a multi-scale fusion process [7] we highlight regions of high contrast, saliency and color saturation on the final result.…”

“…The performance of L 2 UWE is compared to that of five underwater-specific image enhancers [10,16,19,20,39], as well as two low-light-specific enhancers [22,49] using the OceanDark dataset [39]. Seven distinct metrics (i.e., UIQM [37], PCQI [45], FADE [17], GCF [34], r and escore [23], SURF features [9]) show that L 2 UWE outperforms the compared methods, achieving a significant en-hancement in overall image visibility (by reducing low-light regions) and emphasizing the image features (e.g., edges, corners, textures).…”

“…Drews et al [19] considered only two color channels when using the dehazing approach of He et al [24], adapting this method to underwater scenes. Marques et al [39] introduced a contrast-based approach inspired in the Dark Channel Prior [24] that significantly improved the quality of low-light underwater images.…”