Image Illumination Research Articles

Real-Time High Dynamic Equalization Industrial Imaging Enhancement Based on Fully Convolutional Network

Severe reflections on the surfaces of smooth objects can result in low dynamic range and uneven illumination in images, which negatively impacts downstream tasks such as defect detection and QR code recognition on images of smooth workpieces. Consequently, this paper proposes a novel approach to real-time high dynamic equalization imaging based on a fully convolutional network, termed Multi-exposure Image Fusion with Multi-dimensional Attention Mechanism and Training Storage Units (MEF-AT). Specifically, this paper innovatively proposes using training storage units, which utilize intermediate results during network training as auxiliary images, to remove uneven illumination and enhance image dynamic range effectively. Furthermore, by integrating a multi-dimensional attention mechanism into the backbone network, the model can more efficiently extract and utilize critical image information. Additionally, this paper introduces a Deep Guided Filter (DGF) with learnable parameters, which upsample the weight maps generated by the network, thus better adapting to complex industrial scenarios and producing higher quality fused images. An image evaluation metric assessing the lighting uniformity is introduced to thoroughly evaluate the proposed method’s performance. Given the lack of an MEF dataset for smooth workpieces, this paper collects a new dataset for multi-exposure fusion tasks on metallic workpieces. Our method takes less than 4 ms to run four 2K images on a GPU 3090. Both qualitative and quantitative experimental results demonstrate our method’s superior comprehensive performance in proprietary industrial and public datasets.

Polarization-controlled structured illumination for high-resolution imaging

Structured illumination microscopy is a powerful technique that has significantly advanced biological imaging by doubling the resolution compared to conventional methods. However, further resolution enhancement in SIM typically requires complex optical configurations that demand precise control of the incident light's polar and azimuthal angles. To address this challenge, we designed polarization-controlled structured illumination that leverages a dipole resonance all-dielectric super-lattice metasurface (ADSLM). This approach, as demonstrated through simulation, enables high-contrast, high-fidelity object reconstruction with over a 4-fold resolution enhancement. Furthermore, the polarization-controlled ADSLM eliminates the need for complex optical alignments and mechanical adjustments, offering significant potential for integrated high-performance applications in biological imaging.

Effects of Longitudinal External Magnetic Field on Metal Transfer Behavior and Spatter Formation in CO2 Arc Welding

Excessive spatter formation in conventional CO2 arc welding significantly diminishes welding quality and efficiency, posing a critical challenge for industrial applications. To address this issue, this study investigated the mechanisms of metal transfer behavior and spatter formation under the influence of a longitudinal magnetic field (LMF) using a shadow-graph technique with high-speed imaging and back-laser illumination, also coupled with Computational Fluid Dynamics (CFD)-based arc-droplet numerical simulations. The results show that increasing the magnetic flux density (MFD) from 0 to 2 mT shifted the transfer mode from the repelled transfer to the globular transfer, while higher MFDs (3–4 mT) induced rotating repelled transfer. The globular transfer at 2 mT was considered to be primarily produced by the centrifugal effect due to the rotational motion of the molten metal inside the droplet, which was caused by the Lorentz force affected by LMF. The higher droplet temperature in this condition also contributed to forming this transfer mode, preventing the formation of repelled transfer through a decrease in the arc pressure. On the contrary, in the higher MFDs, the droplet temperature decreased to increase the arc pressure, lifting the droplet up. Furthermore, the very strong centrifugal effect rotated the molten metal column around the wire axis to induce the rotating repelled transfer. The spatter formation was found to occur with the two-stage motion of the curved long tail without LMF and at 4 mT, and also with the exploding molten metal column at 4 mT, due to an imbalance of the Lorentz force acting on the molten metal. On the other hand, the neck formation facilitated smooth droplet detachment without forming the curved long tail at 2 mT, reducing spatter significantly. These findings offer valuable insights for optimizing welding quality and efficiency by stabilizing globular transfer under an optimal LMF.

Dark-field X-ray microscopy with structured illumination for three-dimensional imaging

Dark-field X-ray microscopy is a lens-based technique that enables real-space imaging of heterogeneous micro- and meso-scale ordered materials. However, achieving accurate three-dimensional (3D) reconstruction often requires meticulous sample alignment or rastering, requiring complex rotational setups and extended acquisition times. To address these challenges, we introduce a structured illumination technique optimized for 3D imaging of ordered materials at sub-micrometer length scales. Our approach employs a coded aperture to spatially modulate the incident X-ray beam, enabling 3D structural reconstruction from images captured at various aperture positions. Unlike current 3D imaging approaches, which often rely on rotational or rastering methods, our technique uses scanning X-ray silhouettes of the coded aperture for depth resolution along the diffraction axis. This eliminates the need for sample rotation or rastering, resulting in a highly stable and efficient imaging modality. We validated the efficacy of this approach through experimental imaging of an isolated twin domain within a bulk single crystal of an iron pnictide using a dark-field X-ray microscope. This advancement aligns with the enhanced brightness upgrades of modern synchrotron radiation facilities, unlocking new possibilities for high-resolution imaging of ordered materials.

When low-light meets flares: Towards Synchronous Flare Removal and Brightness Enhancement.

Low-light image enhancement (LLIE) aims to improve the visibility and illumination of low-light images. However, real-world low-light images are usually accompanied with flares caused by light sources, which make it difficult to discern the content of dark images. In this case, current LLIE and nighttime flare removal methods face challenges in handling these flared low-light images effectively: (1) Flares in dark images will disturb the content of images and cause uneven lighting, potentially resulting in overexposure or chromatic aberration; (2) the slight noise in low-light images may be amplified during the process of enhancement, leading to speckle noise and blur in the enhanced images; (3) the nighttime flare removal methods usually ignore the detailed information in dark regions, which may cause inaccurate representation. To tackle the above challenges yet meaningful problems well, we propose a novel image enhancement task called Flared Low-Light Image Enhancement (FLLIE). We first synthesize several flared low-light datasets as the training/inference data, based on which we develop a novel Fourier transform-based deep FLLIE network termed Synchronous Flare Removal and Brightness Enhancement (SFRBE). Specifically, a Residual Directional Fourier Block (RDFB) is introduced that learns in the frequency domain to extract accurate global information and capture detailed features from multiple directions. Extensive experiments on three flared low-light datasets and some real flared low-light images demonstrate the effectiveness of SFRBE for FLLIE.

Simulation Analysis of an ETC Monitoring and Imaging Supplementary Lighting Device for Freeways

To address the glare issues caused by highway ETC gantry monitoring and imaging illumination devices, this paper first investigates the mechanism of disability glare and derives the formula for the glare threshold increment. Subsequently, using the glare increment threshold as the evaluation metric and incorporating the current regulatory requirements for illumination devices, a simulation model for ETC gantry monitoring and imaging illumination devices was developed. A single-variable control method was applied to conduct simulation experiments on the glare problems from multiple perspectives (e.g., different standard illuminance levels, various luminous areas, varying installation heights, and different lateral offsets), and the glare level was analyzed using the glare increment threshold method. It was found that when the lateral offset of the illumination device reached 4 m, the glare increment threshold decreased by more than 50%. Additionally, it is recommended that the illuminance of the illumination device should be greater than 15 lx.

Multi-Scale Long- and Short-Range Structure Aggregation Learning for Low-Illumination Remote Sensing Imagery Enhancement

Profiting from the surprising non-linear expressive capacity, deep convolutional neural networks have inspired lots of progress in low illumination (LI) remote sensing image enhancement. The key lies in sufficiently exploiting both the specific long-range (e.g., non-local similarity) and short-range (e.g., local continuity) structures distributed across different scales of each input LI image to build an appropriate deep mapping function from the LI images to their corresponding high-quality counterparts. However, most existing methods can only individually exploit the general long-range or short-range structures shared across most images at a single scale, thus limiting their generalization performance in challenging cases. We propose a multi-scale long–short range structure aggregation learning network for remote sensing imagery enhancement. It features flexible architecture for exploiting features at different scales of the input low illumination (LI) image, with branches including a short-range structure learning module and a long-range structure learning module. These modules extract and combine structural details from the input image at different scales and cast them into pixel-wise scale factors to enhance the image at a finer granularity. The network sufficiently leverages the specific long-range and short-range structures of the input LI image for superior enhancement performance, as demonstrated by extensive experiments on both synthetic and real datasets.

Deep Learning-Based Pointer Meter Reading Recognition for Advancing Manufacturing Digital Transformation Research.

With the digital transformation of the manufacturing industry, data monitoring and collecting in the manufacturing process become essential. Pointer meter reading recognition (PMRR) is a key element in data monitoring throughout the manufacturing process. However, existing PMRR methods have low accuracy and insufficient robustness due to issues such as blur, uneven illumination, tilt, and complex backgrounds in meter images. To address these challenges, we propose an end-to-end PMRR method based on a decoupled circle head detection algorithm (YOLOX-DC) and a Unet-like pure Transformer segmentation network (PM-SwinUnet). First, according to the characteristics of the pointer dial, the YOLOX-DC detection algorithm is designed based on the exceeding you only look once detector (YOLOX). The decoupled circle head of YOLOX-DC detects the pointer meter dial more accurately than the commonly used rectangular detection head. Second, the window multi-head attention of the PM-SwinUnet network enhances the feature extraction ability of pointer meter images and solves problems of missed scale detection and incomplete pointer segmentation. Additionally, the scale and pointer fitting module is introduced into the PM-SwinUnet to locate the accurate position of the scale and pointer. Finally, through the angle relationship between the pointer and the first two main scale lines, the pointer meter reading is accurately calculated by the improved angle method. Experimental results demonstrate the effectiveness and superiority of the proposed end-to-end method across three-pointer meter datasets. Furthermore, it provides a rapid and robust approach to the digital transformation of manufacturing systems.

Quality assurance of hyperspectral imaging systems for neural network supported plant phenotyping

BackgroundThis research proposes an easy to apply quality assurance pipeline for hyperspectral imaging (HSI) systems used for plant phenotyping. Furthermore, a concept for the analysis of quality assured hyperspectral images to investigate plant disease progress is proposed. The quality assurance was applied to a handheld line scanning HSI-system consisting of evaluating spatial and spectral quality parameters as well as the integrated illumination. To test the spatial accuracy at different working distances, the sine-wave-based spatial frequency response (s-SFR) was analysed. The spectral accuracy was assessed by calculating the correlation of calibration-material measurements between the HSI-system and a non-imaging spectrometer. Additionally, different illumination systems were evaluated by analysing the spectral response of sugar beet canopies. As a use case, time series HSI measurements of sugar beet plants infested with Cercospora leaf spot (CLS) were performed to estimate the disease severity using convolutional neural network (CNN) supported data analysis.ResultsThe measurements of the calibration material were highly correlated with those of the non-imaging spectrometer (r>0.99). The resolution limit was narrowly missed at each of the tested working distances. Slight sharpness differences within individual images could be detected. The use of the integrated LED illumination for HSI can cause a distortion of the spectral response at 677nm and 752nm. The performance for CLS diseased pixel detection of the established CNN was sufficient to estimate a reliable disease severity progression from quality assured hyperspectral measurements with external illumination.ConclusionThe quality assurance pipeline was successfully applied to evaluate a handheld HSI-system. The s-SFR analysis is a valuable method for assessing the spatial accuracy of HSI-systems. Comparing measurements between HSI-systems and a non-imaging spectrometer can provide reliable results on the spectral accuracy of the tested system. This research emphasizes the importance of evenly distributed diffuse illumination for HSI. Although the tested system showed shortcomings in image resolution, sharpness, and illumination, the high spectral accuracy of the tested HSI-system, supported by external illumination, enabled the establishment of a neural network-based concept to determine the severity and progression of CLS. The data driven quality assurance pipeline can be easily applied to any other HSI-system to ensure high quality HSI.

Method for simultaneous reconstruction of a depth and clear image with a single blurred image in microscopy

The evaluation of imaging blur degradation characteristics of high-magnification optical microscopes is greatly influenced by complex imaging mechanisms, image textures, and illumination, which seriously limit the observation precision at the micro-nano scale. This paper proposes a method for simultaneous reconstruction of the depth and clear image of a blurred image based on the light intensity distribution law of the microscopic imaging system. First, based on the diffraction characteristics of the light in the circular stable cavity, the light intensity distribution function on the imaging plane of the imaging system is established, and the law of the light intensity diffusion degree with the scene depth variation is obtained by curve fitting, that is, the 3D blur degradation model of the system. Secondly, the normalized blurring degree of blurred images with different textures and different illuminations is calculated, and the mapping relationship between the blurring degree of different images and the light intensity diffusion degree of the system is established with the depth change as the intermediate variable. Thirdly, an adaptive spectral clustering method is introduced to classify the blurred images, and the weighted K-nearest neighbor method is used to automatically classify any blurred image and calculate its normalized blurring degree value and the corresponding system energy diffusion value. Based on the 3D blur degradation model and the normalized blurring degree, the depth calculation of the blurred image and the reconstruction of the clear image are realized simultaneously. The precision of the method proposed in this paper is verified by various standard nano-scale grid images and various real biological tissue samples.

Monocular Vision-Based 3D Reconstruction Method for Parts in Opto-Mechanical Assembly

Abstract In automated assembly, accurately capturing multi-dimensional parameters of parts is essential. Traditional methods include teaching methods or laser-based approaches, with the former relying on experience for pre-calibration and the latter being costly. This paper proposes a low-cost, monocular vision-based method for 3D reconstruction in opto-mechanical assembly. Taking the laser crystal component, including the laser gain medium and its heat sink, in a solid-state laser as an example, a monocular camera is firstly mounted on a robot and captures images of parts from specific angles. Due to the reflective properties of metal surfaces, some image information is lost. To address this problem, a 2D gamma function-based adaptive illumination correction algorithm is then proposed. Next, the improved SIFT algorithm is used to extract and match feature points from the collected images. Subsequently, the SFM approach generates a sparse point cloud, which is then densified using the CMVS/PMVS algorithm. Finally, the Poisson surface reconstruction algorithm is used to complete the 3D reconstruction of the laser crystal heat sinks. Experimental results indicate that the adaptive correction algorithm for non-uniform illumination images can extract more feature points and improve the number of matches. The reconstructed 3D model has high accuracy, providing important technical support for the next step in achieving automated opto-mechanical assembly.

Low illumination image enhancement based on improved CLAHE algorithm

Abstract In the case of low light, the target features will be weakened or even disappear, thus affecting the target detection effect based on aerial images. In order to improve the image effect, an improved CLAHE algorithm is designed. In the process of image enhancement, the CLAHE algorithm is improved according to image pixel size and histogram characteristics, which can improve the smoothness of CLAHE algorithm for uniform color parts. The experimental results show that the improved CLAHE algorithm is better than other histogram equalization algorithms, and the method has low cost and practical application value.

Information in the Shroud of Turin About Its Variable Molecular Properties

Three earlier papers on the Shroud of Turin showed that ultraviolet induced fluorescence (UVIF) intensity is non-uniform over the surface of the Shroud. These intensity results were based on an analysis of UVIF photographic images taken in 1978 as part of a scientific investigation of the Shroud. In this paper the statistical technique, principal component analysis (PCA), is used to analyze the information present in the UVIF images. It is shown that the vast majority of information in the UVIF Shroud images is contained in their International Commission on Illumination (CIE Lab) image intensity. Differences in UVIF intensity indicate differences in the Shroud’s molecular properties. Such differences can be due to differences in molecular structure, contamination, and burned regions. Fluorescence spectral results published as part of the same 1978 scientific study confirm the intensity variation with position of the UVIF images. These spectral fluorescence results are discussed in detail. The combined photographs and spectral results demonstrate that color contributes relatively little to the information in the UVIF images. One possible cause of the variation of UVIF intensity, namely neutron radiation, is briefly discussed. Charred Shroud material, collected during its cleaning in 2002, can be used to assess whether the Shroud was exposed to neutron radiation. Such testing would be completely non-destructive to the Shroud. New UVIF photographs need to be taken to validate the results discussed in this paper.

Optimized galvanometric illumination for terahertz full-field imaging and computed tomography

The pursuit of high-resolution, high-fidelity, real-time imaging is receiving significant attention in terahertz community. In this study, a versatile illumination approach based on a double-mirror galvanometer is proposed and optimized for multiple terahertz imaging approaches. We analyzed the mechanism of galvanometric illumination and elucidated two main factors affecting its homogeneity and parallelism properties. In our module, the terahertz beam is deflected rapidly by the galvanometer which is driven by triangular voltage signals, and then focused by a self-designed aspherical f-θ lens to illuminate the object at an equal lateral scanning velocity. The object beam of different transients is periodically superimposed along the Lissajous trajectory, and is recorded by an array microbolometer in a single integration of 3 s. A homogeneous illumination field is realized with a speckle contrast of 0.11, and the resultant image achieves isotropic resolution. By adopting the proposed galvanometric illumination strategy, the average intensity of the illumination field is increased by 135% compared to expanded coherent illumination, and the speckle contrast is reduced by 83.5% compared to other galvanometric illumination. By virtue of leading low speckle noise, illumination homogeneity and parallelism, a compact imaging system is built for terahertz full-field imaging and computed tomography achieving high imaging speed and fidelity. As a successful attestation of terahertz beam steering, this study is very promising to reduce the total cost, increase the performance and expand the application of terahertz imaging.

Fluorescence and color adjustment potentials of paste-type and flowable resin composites in cervical restorations.

Comparatively assess the fluorescence and color adjustment potentials of paste-type and flowable resin composites performed in cervical restorations. Five paste-type resin composites and 5 highly-filled flowable composites were investigated (n=5 for each). Class V cavities on 50 extracted A2 shade human central incisors were restored. Cross-polarization (CP) and fluorescence illumination (FI) images were recorded for each restoration. L*a*b* values were analyzed using the Photoshop software and color adjustment (ΔECP) and fluorescence adjustment (ΔEFI) levels were quantitatively assessed. Kruskal Wallis Test, Independent Samples t Test, One Way Anova-Welch test, and Pearson Correlation were used for the statistical analyses. The deemed significance was set at p<.050. Paste-type composites presented significantly lower ΔEFI and ΔECP values than the flowable composites (p=.006 and p<.001 respectively). ΔEFI and ΔECP values were positively correlated (r=0.584; p<.001). Paste-type Gaenial A'chord presented the lowest ΔEFI value [7.67 (5.35 - 10.99)b], while Estelite Asteria presented significantly the highest [19.94 (17.08 - 22.17)c]. Charisma Diamond One Flow presented the highest ΔECP value [3.25 (3.13 - 3.39)a] and Gaenial A'chord presented the lowest [0.79 (0.48 - 0.87)e] which was the only imperceptible score (<0.8). The only clinically acceptable color adjustment was found for Geanial Universal Injectable [1.06 (0.73 - 1.62)e] among the flowable composites. ΔEFI and b* values were positively correlated (r=0.493; p<.001). Color adjustment level and fluorescence adjustment level were considered coherent for particular composite brands. At the same time, both were better for the paste-type composites than the flowable composites on A2 base shade class V restorations. Fluorescence adjustment potential as well as the color adjustment potential varied among different composite brands regarding paste-type and flowable composites. Understanding both the color and fluorescence adjustment potentials of paste-type and flowable resin-based composite materials is clinically vital to achieve the high-end esthetics.

Unifying domain gap in Federated Learning: A geometric approach

Federated Learning (FL) is a framework designed to enable distributed and privacy-preserving learning. It typically involves a central server coordinating multiple clients to collaboratively train a global model without the need for sharing private data. Recent studies have revealed that this parameter-only communication strategy can be hindered by non-independent and identically distributed (non-iid) data across clients, leading to statistical heterogeneity. This issue necessitates the development of Personalized Federated Learning (PFL), which aims to enhance the performance of individual clients while still allowing for the learning of a global model. Most existing work on PFL has focused on label shift in supervised tasks, with relatively little attention given to feature shift, another common problem in real-world applications of federated learning. For instance, client data may vary in terms of illumination, angles, and quality for natural images, or in imaging protocols for medical images. To address this issue, we propose a general yet powerful framework that leverages the geometric interpretation of deep neural networks. Geometrically, a deep neural network partitions the feature space into a Power Diagram (PD) or Voronoi Diagram (VD). Regularized losses are introduced to mitigate feature shift from the perspective of PD and VD. Additionally, a personalized generator network is formulated for each client to rectify feature distribution and adapt the input data from its own domain to an ”averaged” or unified domain that generalizes well across all clients. Our method demonstrates competitive performance compared to current state-of-the-art methods.

Stochastically structured illumination microscopy scan less super resolution imaging

In super-resolution, a varying illumination image stack is required. This enriched dataset typically necessitates precise mechanical control and micron-scale optical alignment and repeatability. Here, we introduce a novel methodology for super-resolution microscopy called stochastically structured illumination microscopy (S2IM), which bypasses the need for illumination control exploiting instead the random, uncontrolled movement of the target object. We tested our methodology within the clinically relevant ophthalmoscopic setting, harnessing the inherent saccadic motion of the eye to induce stochastic displacement of the illumination pattern on the retina. We opted to avoid human subjects by utilizing a phantom eye model featuring a retina composed of human induced pluripotent stem cells (iPSC) retinal neurons and replicating the ocular saccadic movements by custom actuators. Our findings demonstrate that S2IM unlocks scan-less super-resolution with a resolution enhancement of 1.91, with promising prospects also beyond ophthalmoscopy applications such as active matter or atmospheric/astronomical observation.

Phase changes in burning precursor-laden single droplets leading to puffing and micro-explosion

When producing metal-oxide nanoparticles via flame spray pyrolysis, precursor-laden droplets are ignited and undergo thermally induced disintegration, called ‘puffing’ and ‘micro-explosion’. In a manner that is not fully understood, these processes are associated with the formation of dispersed phases inside the droplets. This work aims at visualizing the interior of precursor-laden burning single droplets via diffuse back illumination and microscopic high-speed imaging. Solutions containing iron(III) nitrate nonahydrate (INN) and tin(II) 2-ethylhexanoate (Sn-EH) were dispersed into single droplets of sub-100 μm diameter that were ignited by passing through a heated coil. At low precursor concentration, 50% of the INN-laden droplets indicate a gas bubble of about 5 μm diameter in the center of the droplet. The bubble persists for several hundred microseconds at a similar size. In almost all of these cases, the bubble expands at some point and the droplet ends up in a micro-explosion. In some of these instances, the droplet’s surface shows spatial brightness modulations, i.e., surface undulations, indicating the formation of a viscous shell. With increasing INN concentration, the fraction of droplets showing surface undulations, gas bubbles, and micro-explosions drastically decreases. This may be associated with a more rigid viscous shell and reduced mobility of bubbles. Bright incandescent streaks originating from the disrupting INN-laden droplets, may indicate sub-micrometer droplets or particles from within the droplets or formed in the gas phase. In contrast, Sn-EH-laden droplets show very fast disruptions, typically less than 10 μs from first visible deformation to ejection of secondary droplets. Bubbles and surface undulations were not observed.Graphical abstract

An Objective Evaluation Method for Image Sharpness Under Different Illumination Imaging Conditions

Blurriness is troublesome in digital images when captured under different illumination imaging conditions. To obtain an accurate blurred image quality assessment (IQA), a machine learning-based objective evaluation method for image sharpness under different illumination imaging conditions is proposed. In this method, the visual saliency, color difference, and gradient information are selected as the image features, and the relevant feature information of these three aspects is extracted from the image as the feature value for the blurred image evaluation under different illumination imaging conditions. Then, a particle swarm optimization-based general regression neural network (PSO-GRNN) is established to train the above extracted feature values, and the final blurred image evaluation result is determined. The proposed method was validated based on three databases, i.e., BID, CID2013, and CLIVE, which contain real blurred images under different illumination imaging conditions. The experimental results showed that the proposed method has good performance in evaluating the quality of images under different imaging conditions.

Zero-UMSIE: a zero-shot underwater multi-scale image enhancement method based on isomorphic features.

Due to the scattering and absorption of light, underwater images often exhibit degradation. Given the scarcity of paired real-world data and the inability of synthetic paired data to perfectly approximate real-world data, it's a challenge to restore these degraded images using deep neural networks. In this paper, a zero-shot underwater multi-scale image enhancement method (Zero-UMSIE) is proposed, which utilizes the isomorphism between the original underwater image and the re-degraded image. Specifically, Zero-UMSIE first estimates three latent components of the original underwater image: the global background light, the transmission map, and the scene radiance. Then, the estimated scene radiance is randomly mixed with the original underwater image to generate re-degraded images. Finally, a multi-scale loss and a set of tailored non-reference loss functions are employed to fine-tune the underwater image and enhance the generalization ability of the network. These functions implicitly control the learning preferences of the network and effectively address issues such as color bias and uneven illumination in underwater images, without the need for additional datasets. The proposed method is evaluated on three widely used real-world underwater image datasets. Extensive experiments on various benchmarks demonstrate that the proposed method is superior to state-of-the-art methods subjectively and objectively, which is competitive and applicable to diverse underwater conditions.