Residual Haze Research Articles

Spatial Variance of Fusion and Physics-Aware for Single Image Dehazing

Currently, most image dehazing algorithms overlook the local details of the image and fail to fully exploit different levels of features, resulting in color distortion, decreased contrast, and residual haze in the restored haze-free images. To tackle these issues, this paper proposes a method that dynamically enhances pixels by utilizing the spatial variation of image dark channel prior in images. For regions with high haze density where local information is insufficient, a Transformer is employed to learn the global dependencies of the input features. Conversely, for regions with low haze density where local information is effective, parallel multi-scale attention is used to extract local features. When enhancing each pixel, we dynamically determine the contribution of non-local and local information based on the image features. Furthermore, to better reflect the physical process of haze formation in the image and improve the interpretability of the feature space, a dual-branch physics-aware unit is established. It learns the features related to atmospheric scattering in the image and captures the visual characteristics. In the experiments, a large dataset of dehazing images is used for training and testing, and comparisons are made with other existing dehazing methods. The results demonstrate that this method, which takes into account both local and global information, significantly improves the dehazing performance of the model.

PSD-ELGAN: A pseudo self-distillation based CycleGAN with enhanced local adversarial interaction for single image dehazing

Compared to pixel-level content loss, domain-level style loss in CycleGAN-based dehazing algorithms just imposes relatively soft constraints on the intermediate translated images, resulting in struggling to accurately model haze-free features from real hazy scenes. Furthermore, globally perceptual discriminator may misclassify real hazy images with significant scene depth variations as clean style, thereby resulting in severe haze residue. To address these issues, we propose a pseudo self-distillation based CycleGAN with enhanced local adversarial interaction for image dehazing, termed as PSD-ELGAN. On the one hand, we leverage the characteristic of CycleGAN to generate pseudo image pairs during training. Knowledge distillation is employed in this unsupervised framework to transfer the informative high-quality features from the self-reconstruction network of real clean images to the dehazing generator of paired pseudo hazy images, which effectively improves its haze-free feature representation ability without increasing network parameters. On the other hand, in the output of dehazing generator, four non-uniform image patches severely affected by residual haze are adaptively selected as input samples. The local discriminator could easily distinguish their hazy style, thereby further compelling the dehazing generator to suppress haze residues in such regions, thus enhancing its dehazing performance. Extensive experiments show that our PSD-ELGAN can achieve promising results and better generality across various datasets.

Dual-stream complex-valued convolutional network for authentic dehazed image quality assessment.

Effectively evaluating the perceptual quality of dehazed images remains an under-explored research issue. In this paper, we propose a no-reference complex-valued convolutional neural network (CV-CNN) model to conduct automatic dehazed image quality evaluation. Specifically, a novel CV-CNN is employed that exploits the advantages of complex-valued representations, achieving better generalization capability on perceptual feature learning than real-valued ones. To learn more discriminative features to analyze the perceptual quality of dehazed images, we design a dual-stream CV-CNN architecture. The dual-stream model comprises a distortion-sensitive stream that operates on the dehazed RGB image, and a haze-aware stream on a novel dark channel difference image. The distortion-sensitive stream accounts for perceptual distortion artifacts, while the haze-aware stream addresses the possible presence of residual haze. Experimental results on three publicly available dehazed image quality assessment (DQA) databases demonstrate the effectiveness and generalization of our proposed CV-CNN DQA model as compared to state-of-the-art no-reference image quality assessment algorithms.

TCRN: A Two-Step Underwater Image Enhancement Network Based on Triple-Color Space Feature Reconstruction

The underwater images acquired by marine detectors inevitably suffer from quality degradation due to color distortion and the haze effect. Traditional methods are ineffective in removing haze, resulting in the residual haze being intensified during color correction and contrast enhancement operations. Recently, deep-learning-based approaches have achieved greatly improved performance. However, most existing networks focus on the characteristics of the RGB color space, while ignoring factors such as saturation and hue, which are more important to the human visual system. Considering the above research, we propose a two-step triple-color space feature fusion and reconstruction network (TCRN) for underwater image enhancement. Briefly, in the first step, we extract LAB, HSV, and RGB feature maps of the image via a parallel U-net-like network and introduce a dense pixel attention module (DPM) to filter the haze noise of the feature maps. In the second step, we first propose the utilization of fully connected layers to enhance the long-term dependence between high-dimensional features of different color spaces; then, a group structure is used to reconstruct specific spacial features. When applied to the UFO dataset, our method improved PSNR by 0.21% and SSIM by 0.1%, compared with the second-best method. Numerous experiments have shown that our TCRN brings competitive results compared with state-of-the-art methods in both qualitative and quantitative analyses.

Intra-stromal injection of honey-treated keratocytes as a cell-based therapy for experimental corneal laceration.

This study aimed to investigate the potential of honey-supplemented medium (HSM) for expanding corneal keratocytes and its transplantation in a model of corneal laceration. Keratocytes were cultured in 1 % HSM- or 10 % fetal bovine serum (FBS)-supplemented medium for 24 h. The effect of HSM on keratocyte proliferation was evaluated using the MTT assay. The relative expression of Lum, Kera, and ALDH3A1, known markers of native keratocytes, was quantified by real-time PCR. The safety and efficacy of HSM-treated keratocyte intrastromal injection in a rabbit model of corneal laceration were also evaluated. The MTT assay showed that HSM treatment didnot significantly affect cell viability compared to FBS-supplemented medium (84.71±2.38 vs. 100.08±10.92, respectively; p=0.076). Moreover, HSM-treated keratocytes had significantly increased expression of Lum, Kera,andALDH3A1 compared to cells treated with FBS, while theexpression of the proliferation biomarker Thy-1 did notsignificantly differ between the two treatments. Intrastromal injection of HSM-treated keratocytes in the laceration animal model was safe and uneventful, resulting in less stromal inflammation and neovascularization, and consequently, better final architecture with less residual haze compared to the group injected with FBS-treated keratocytes. These findings suggest that honey is a suitable supplement for keratocyte treatment and corneal cell therapy. The use of HSM may have potential applications in the treatment of corneal injuries and diseases.

Single image haze removal via attention-based transmission estimation and classification fusion network

With the rapid development of deep learning theory, many excellent convolutional neural network (CNN) based dehazing methods have been proposed for single image dehazing. However, the slow training convergence rate and haze residual are still two serious flaws of these existing dehazing networks. To tackle these issues, we propose a novel end-to-end CNN-based dehazing framework called attention-based transmission estimation and classification fusion network (ATECFN). The ATECFN framework consists of three submodules: attention-based transmission-airlight estimation network (ATAEN), multi-scale Auto-Encoders (MAE), and patch-based classification fusion network (PCFN). First, the transmission similarity, that is the similarity of neighboring pixels in transmission map, is introduced to significantly increase the capability of CNN to fit transmission map. Second, the ATAEN is exploited to estimate airlight map and transmission map and then they are used to obtain a rough dehazed result according to the atmospheric scattering model. Third, we present the MAE to further refine the rough result, where the multi-scale structure can effectively capture local details over a wide range of scales. Finally, PCFN, a new fusion strategy, is employed to integrate the two results generated by ATAEN and MAE, in which a probability map is derived from a binary classification network and viewed as the fusion coefficient map. Extensive experiments demonstrate that the proposed algorithm outperforms the state-of-the-art methods on both synthetic and real-world images, which can not only improve the training convergence rate but also remove the residual haze effectively.

Low-driving-voltage, polarizer-free, scattering-controllable liquid crystal device based on randomly patterned photo-alignment.

In this work, a polarizer-free, electrically tunable liquid crystal device is presented. This device is based on randomly patterned photo-aligned boundaries generating light scattering in an adjacent liquid crystal film. Its transparent on state requires only 7.5Vrms driving voltage. Switching from a 49.5% hazy off state into the transparent on state occurs at 1.25 V, with only 1.2% residual haze. The effect exhibits fast rise and decay times of 0.3 ms and 7.2 ms, respectively. Thanks to its field-effect nature and the surface preparation process, zero ohmic low power consumption, fast response times, and steep transmission-voltage characteristics result. Applications are smart windows, light shutters, and/or transparent displays, as well as complementary metal-oxide-semiconductor driven multiplexed displays.

Deep Residual Haze Network for Image Dehazing and Deraining

Image dehazing on a hazy image aims to remove the haze and make the image scene clear, which attracts more and more research interests in recent years. Most existing image dehazing methods use a classic atmospheric scattering model and natural image priors to remove the image haze. In this paper, we propose an end-to-end image dehazing model termed as DRHNet (Deep Residual Haze Network), which restores the haze-free image by subtracting the learned negative residual map from the hazy image. Specifically, DRHNet proposes a context-aware feature extraction module to aggregate the contextual information effectively. Furthermore, it proposes a novel nonlinear activation function termed as RPReLU (Reverse Parametric Rectified Linear Unit) to improve its representation ability and to accelerate its convergence. Extensive experiments demonstrate that DRHNet outperforms state-of-the-art methods both quantitatively and qualitatively. In addition, experiments on image deraining task show that DRHNet can also serve for image deraining.

Subtotal Transepithelial Phototherapeutic Keratectomy with Mitomycin C Application for Central Corneal Opacities-Three Case Series

Purpose: To evaluate the beneficial effects of transepithelial subtotal phototherapeutic keratectomy (PTK) with mitomycin C application for corneal opacities on patients’ visual acuity. Methods: Three cases with central corneal opacities were subjected to transepithelial subtotal (PTK) using Excimer laser ablation with mitomycin C (0.02%) application for 2 minutes, the first case with a central post-PRK with mitomycin corneal scar 98 µ depth, where PTK ablation depth only 70 µ (thin cornea=430 µ), the second case with a central post-traumatic corneal scar 121 µ depth, with PTK ablation depth only 70 µ (thin cornea=480 µ) and the third case with right corneal (granular) dystrophy, with opacity level was 200 µ, where PTK of only 100 µ was done (young female patient). Results: The BCVA increased from 0.2 to 0.7 in case 1 and from 0.2 to 0.5 in case 2 and from 0.05 to 0.5 in case 3, without evidence of residual haze in case 1, and with minimal residual haze in case 2 and case 3. Conclusion: In the presence of relatively non-superficial corneal opacities of various causes, subtotal transepithelial phototherapeutic keratectomy with mitomycin C application can result in marked improvement of visual acuity without the need for keratoplasty.

Corneal Haze Determined by Confocal Microscopy 2 Years After Descemet Stripping With Endothelial Keratoplasty for Fuchs Corneal Dystrophy

To quantify corneal light scatter and its relationship to vision after Descemet stripping with endothelial keratoplasty (DSEK). Eyes with Fuchs corneal dystrophy were examined before and 1, 3, 6, 12, and 24 months after DSEK. Outcome measures were high- and low-contrast visual acuity, contrast sensitivity, and forward light scatter. Corneal reflectivity (backscatter), expressed in scatter units (SU), was quantified using in vivo confocal microscopy. Comparing 49 study eyes with 35 normal eyes, the mean (SD) corneal subepithelial layer was more reflective than normal before (2325 [613] vs 1208 [287] SU, P<.001) and through 24 months after DSEK (1760 [432] SU, P<.001). Interface reflectivity remained higher than in normal stroma through 24 months (1228 [287] vs 827 [188] SU, P<.001). At 1 year, forward light scatter correlated with subepithelial reflectivity (r=0.28, P=.01) but not interface reflectivity. Recipient age was correlated with improvement in subepithelial reflectivity at 12 months (r=–0.41, P=.01, 34 eyes) and 24 months (r=–0.36, P=.02, 26 eyes) after DSEK, and the improvement of subepithelial haze in eyes of participants aged 62 years or younger was correlated with improvement in forward light scatter at 12 months (r=0.52, P=.008) and 24 months (r=0.62, P=.004). In Fuchs corneal dystrophy, the corneal subepithelial region is a more important source of forward light scatter than the DSEK interface. Subepithelial haze improves more in younger patients and is associated with improvement in forward light scatter. Visual function after DSEK is affected by residual haze in the anterior host cornea more so than the surgical interface. Haze, which likely is experienced as glare disability, improves after surgical intervention but improves more in younger patients.

Comparative Evaluation of Big-Bubble Deep Anterior Lamellar Keratoplasty and Penetrating Keratoplasty in a Case of Macular Corneal Dystrophy

A case of a 54-year-old patient with macular corneal dystrophy who underwent a penetrating keratoplasty (PK) in the left eye and a deep anterior lamellar keratoplasty (DALK) in the right eye is described. The merits of PK versus DALK for visual rehabilitation in macular corneal dystrophy are discussed. After deep lamellar dissection in the left eye, the remaining bed of residual stroma, Descemet membrane, and endothelium was unacceptably hazy and thus converted to PK. The right eye had successful DALK because a relatively clear bed was noted after deep lamellar separation using the big-bubble technique. The spectacle-corrected visual acuities at 2-year follow-up are 6/6 OD and 6/9+ OS. There is mild residual haze in the right eye, although the visual acuity is better in this eye. Endothelial cell counts were satisfactory and not significantly different in both eyes. This case demonstrates that DALK with the big-bubble technique can be successfully carried out for visual rehabilitation in macular dystrophy. DALK should be considered in cases with no significant Descemet membrane and endothelial involvement because of the potential advantages offered by retaining the host endothelium.

Microbial keratitis after corneal collagen crosslinking

Several infiltrates appeared in the upper midperipheral cornea of a 29-year-old woman who had had uneventful corneal collagen crosslinking (CXL) with riboflavin and ultraviolet-A light (UVA) for the treatment of keratoconus in the right eye. Staphylococcus epidermidis keratitis was confirmed by microbiological studies, which guided treatment with topical fortified antibiotic agents. Before CXL, the best spectacle-corrected visual acuity (BSCVA) in the right eye was 20/25, the manifest refraction was -0.25 -0.25 x 125, and the anterior segment was normal under biomicroscopy. Five months after the procedure, the BSCVA was 20/22, the manifest refraction was +1.00 -2.50 x 40, and slitlamp examination revealed a mild residual haze in the upper midperipheral cornea. Collagen crosslinking with riboflavin-UVA is a minimally invasive method but traditionally requires epithelial removal, which could be a predisposing factor to bacterial keratitis.

Phototherapeutic keratectomy with mitomycin C for corneal haze following photorefractive keratectomy for myopia.

To evaluate the safety and efficacy of phototherapeutic keratectomy (PTK) with single application of mitomycin C for patients with severe corneal haze following photorefractive keratectomy (PRK) for high myopia. Eight eyes of seven patients were treated with PTK and intraoperative topical application of mitomycin C (0.02%) for severe corneal haze (grade 3) following PRK for myopia. All patients' visual performance improved significantly. Mean preoperative visual acuity (20/200 for both UCVA and BSCVA) improved significantly to 20/33 (0.6) and 20/30 (0.7) for UCVA and BSCVA, respectively. Six eyes (85.7%) had improved UCVA to 20/40 or better and gained five or more lines of UCVA. The corneal haze score decreased from grade 3 initially (for all eyes prior to PTK and mitomycin C) to a final mean haze score of 0.3 (range 0 to 0.5). Mean final spherical equivalent refraction achieved was -1.30 +/- 1.60 D (range -3.75 to +1.25 D). One eye gained only three lines of visual acuity due to regression and residual haze. No adverse effects related to the use of mitomycin C were recorded. PTK with a single intraoperative application of mitomycin C was safe and effective in reducing corneal haze and improving visual acuity in patients with severe corneal haze following PRK.

Vertebral angiography in the diagnosis of meningioma within the lateral ventricle.

Reports of meningioma within the lateral ventricle demonstrated by vertebral angiography are few in number. An extensive search through the literature has revealed only five cases in all. Ameli (1952) described one case and published a lateral radiograph demonstrating a tumour circulation. Wall (1954) reviewing a series of eight cases of intraventricular meningioma stated “vertebral angiography which was done in four cases showed in the lateral views one constant feature, namely, absence of the normal pattern of the choroid plexus … In addition, the posterior choroidal vessels were abnormally numerous or stretched. In one instance the tumour was outlined by encircling vessels and a residual haze. The antero-posterior films, as would be expected, were of little help because in this view the choroidal arteries are foreshortened.” Because a tumour circulation with residual “blush” has been described in only two of the cases in the literature, and the antero-posterior vertebral angiographic appearances do not appear to have been previously published, it is felt that the following case of intraventricular meningioma diagnosed by vertebral angiography should be recorded. The patient, a woman of 47, was admitted to Astley Hospital, Leigh, in August, 1959, under the care of Dr. T. B. S. Dick, with the complaint that for five years she had suffered from pains in the back of the neck, and for three months from “pins and needles” in the left cheek.