Underwater images typically exhibit color distortion and poor visibility due to light absorption and scattering. Currently, existing methods always overcompensate for degraded color and contrast due to a lack of adaptation, which results in an unnatural appearance and contrast loss. This article combines the merits of conventional color transfer technology and dehazing to improve underwater image quality while addressing the aforementioned problems. Specifically, a maximum information transfer method that does not require a reference image to adaptively correct the color of an input image is first proposed. Built on maximizing contrast while minimizing contrast loss, an adaptive full dynamic range mapping (AFDRM) strategy is then proposed to guide dehazing to restore the visibility. Our method can produce vivid results without introducing over enhancement and is applicable to a variety of underwater environments. Furthermore, with our sufficient and reasonable proof, our method is extended and applied to low-light image enhancement (LLIE) by fine-tuning parameters in this article. Extensive experiments demonstrate that our method achieves superior color correction and contrast enhancement, as well as remarkable performance in underwater applications and low-light scenes, even for foggy images taken at nighttime and daytime.
Unsupervised low-light image enhancement methods have gained attention and shown improvement with low data dependence. However, the lack of a ground truth presents challenges, notably in pronounced noise and color bias. This paper proposes a Self-Guided Pixel-wise Calibration method to overcome associated issues by leveraging inherent features from the input as a self-guide. Specifically, a Pixel-wise Guided Filter is introduced to decrease noise, utilizing a low-light image for guidance and deep features as regularization maps. Additionally, a Color Correction Module is introduced to enhance saturation by adjusting the shadow threshold. Finally, a pixel-wise exposure control loss is formalized to optimize overall naturalness by adjusting brightness to a well-exposedness map from the low-light image. Extensive experiments demonstrate that our method outperforms many state-of-the-art methods, producing enhanced results with fewer distortions across various real-world image enhancement tasks.
Ship detection from synthetic aperture radar (SAR) images is one of the crucial issues in maritime surveillance. However, it is very difficult to detect ships by traditional detection methods because of the complex background of high-resolution SAR images. We propose a contrast-based target detection method in superpixel level. Experimental results based on ocean SAR images have shown that the proposed method can obtain stable detection performance both in strong clutter and heterogeneous backgrounds. Meanwhile, it has a low computational complexity compared with some existing detection methods. Firstly, the method introduces superpixel segmentation to partition the SAR image. Then according to the difference in the gray distribution and intensity of the target and clutter superpixels, the difference between the target and clutter is increased based on the weighted information entropy and local contrast of the superpixels to achieve target superpixel detection. Finally, the obtained superpixels of the suspected target are clustered, and false alarms are removed based on the geometric characteristics of the target. The method performs detection at the superpixel level, which can ensure the detection accuracy and improve the detection efficiency.