This study proposes a novel low-light image enhancement network that incorporates a frequency cross-attention mechanism in the wavelet domain. The proposed network enhances brightness through the low-frequency wavelet subband while simultaneously restoring fine details in the high-frequency subbands. Color degradation during the brightening process is prevented by applying a color-preserving block based on the saturation component before the illumination adjustment. Furthermore, U-shaped lightening and multiscale sharpening blocks are designed to enhance the image brightness and detail, respectively. The disruption of intrinsic symmetry in coefficient correlations poses a major challenge in independently processing wavelet subbands. To address this issue, we propose a frequency cross-attention block that enables effective information exchange between subbands, thereby preserving their inherent correlations. The proposed network produces visually consistent and refined outputs by balancing the enhanced wavelet subbands. Experimental evaluations demonstrate that the proposed network achieves competitive performance in both subjective quality and objective metrics, confirming its effectiveness for low-light image enhancement.

Underwater image enhancement aims to effectively eliminate the greenish or bluish hues common in underwater environments, thereby improving the performance of computer vision tasks. However, challenges such as severe color distortion and blurring in underwater settings remain significant. This study introduces a two-phase underwater image enhancement network to address these problems. In the first phase, the network uses a green-channel-guided fusion block alongside an adaptive unsharp mask block for each color channel. The green channel, typically less affected by underwater conditions, guides the enhancement of the other channels, particularly the red channel, to reduce color distortion. Furthermore, we introduce a novel trainable unsharp mask block to enhance image contrast, offering a more effective alternative to traditional unsharp masking techniques. In the second phase, the enhanced color channels are further refined to ensure balanced color consistency and improved contrast. To preserve the integrity of the enhanced features, an unsharp mask block is combined with a residual operation. The effectiveness of the proposed network is evaluated through extensive experiments on widely used benchmark datasets, demonstrating that our approach either surpasses or matches the performance of existing state-of-the-art methods for underwater image enhancement. The code is available on https://github.com/Juheon-Iee.

This study proposes a gate-aware recurrent framework for low-light image enhancement. By augmenting the conventional long short-term memory architecture with an additional gating mechanism, a gate-aware recurrent illumination block is proposed to selectively retain and update information across multiple iterations, thereby preserving contextual dependencies while progressively improving illumination. To realize this process, two specialized blocks are incorporated: a gate-aware feature amplification block that selectively enhances brightness-related features through global maximum pooling, and a global illumination refinement block that regulates the overall brightness levels to achieve stable enhancement. Furthermore, initializing hidden states with saturation features from the input low-light image stabilizes color restoration. In addition, an auxiliary refinement step ensures sharper details and more natural brightness reproduction. Extensive experiments, including thorough ablation studies, validate the contributions of each design component. The results demonstrate that the proposed network consistently achieves state-of-the-art performance on both reference-based and no-reference datasets, providing a lightweight yet effective solution for practical low-light image enhancement applications. The code is available on https://github.com/daeun00.

Low-light image enhancement is a challenging problem in computer vision due the inherently fragile nature of low-light images, which suffer from low brightness, poor contrast, and color distortion. This study proposes a contrast-aware low-light image enhancement network in the wavelet domain. The contrast measure defined as the ratio of the high-frequency coefficient to the low-frequency coefficient serves as a key factor in restoring natural illumination. The proposed network includes U-shaped lightening and sharpening blocks, complemented by lightened and sharpened feature attention blocks. The lightening block, which uses wavelet low-frequency subbands as input, is employed to restore brightness. The lightened information is then transmitted to high-frequency subbands through the lightened feature attention block. The sharpening block is used to enhance image detail within high-frequency subbands. The sharpened information is then relayed to the lightened low-frequency subbands to balance the image contrast. The proposed network achieves balanced contrast by facilitating the exchange of enhancement information between the low- and high-frequency subbands. Additionally, we introduce a saturation guide block to restore image color effectively. This study restores each component through wavelet decomposition and uses the concept of contrast to restore it to a natural image. This could be further developed by combining color space decomposition. The effectiveness of the proposed network is evaluated through extensive experiments on well-known benchmarks. Simulation results confirm that the proposed low-light enhancement network surpasses other state-of-the-art approaches.

This paper proposes a machine learning-based underwater image enhancement scheme using an adaptive standardization network and normalization network. The adaptive standardization network is designed to match the distribution of input features. This helps correct the distorted distribution of underwater images and facilitates training. The proposed adaptive normalization network is constructed using two squeeze-and-excitation blocks and the conventional feature normalization method. It is designed to increase the contrast, remove the hazy effect, and restore the brightness. An improved performance of underwater image enhancement is achieved through an appropriate configuration of the two proposed networks. The structure of the proposed network is simple and therefore requires fewer parameters. The simulation results verify that the proposed underwater image enhancement scheme outperforms other state-of-the-art approaches. The proposed method demonstrates outstanding performance both subjectively and objectively in improving underwater images. The code is available on https://github.com/cwoop92.

In this paper, we propose an efficient and fast low-light image enhancement method using an atmospheric scattering model based on an inverted low-light image. The transmission map is derived as a function of two saturations of the original image in the two color spaces. Due to the difficulty in estimating the saturation of the original image, the transmission map is converted into a function of the average and maximum values of the original image. These two values are estimated from a given low-light image using the gamma correction prior. In addition, a pixel-adaptive gamma value determination algorithm is proposed to prevent under- or over-enhancement. The proposed algorithm is fast because it does not require the training or refinement process. The simulation results show that the proposed low-light enhancement scheme outperforms state-of-the-art approaches regarding both computational simplicity and enhancement efficiency. The code is available on https://github.com/TripleJ2543.

This study proposes an edge-preserving image deraining network using a wavelet feature aggregation method. Wavelet subbands re correlated with each other, and the high-frequency subband in the horizontal direction is the least affected by rain streak contamination. On this basis, we introduce a single image deraining network that cumulatively aggregates wavelet subband features according to their importance. The network architecture primarily comprises a wavelet feature aggregation block and a residue channel guide block. The aggregation of features with the cumulative wavelet feature aggregation block moves downward and upward, and a long short-term memory-based multiscale attentive rain streak removal block is developed to serve as the backbone for rain streak removal. We use a residual channel map based on the low-frequency subband to construct guide features that assist in rain streak removal. A repetitive image restoration framework that incorporates two proposed blocks is used to iteratively improve rainy images. We test the proposed network on various image datasets and compare the deraining performance with those of existing methods The experimental results demonstrate that the performance of the proposed scheme is superior to that of other tested deraining methods. https://github.com/SYChoi98/CWFANet/

Low-light images often exhibit reduced brightness, weak contrast, and color distortion. Consequently, enhancing low-light images is essential to make them suitable for computer vision tasks. Nevertheless, addressing this task is particularly challenging because of the inherent constraints posed by low-light environments. In this study, we propose a novel low-light image enhancement network using adaptive feature stretching and informative attention. The proposed network architecture mainly includes an adaptive feature stretch block designed to extend the narrow range of image features to a broader range. To achieve improved image restoration, an informative attention block is introduced to assign weight to the output features from the adaptive feature stretch block. We conduct comprehensive experiments on widely used benchmark datasets to assess the effectiveness of the proposed network. The experimental results show that the proposed low-light image enhancement network yields satisfactory results compared with existing state-of-the-art methods from both subjective and objective perspectives while maintaining acceptable network complexity.

Image rain removal is designed to effectively separate rain streaks from the background image layer. However, rain streaks in real-world scenarios vary in density, shape, and direction, making it difficult to decompose rainy images into clean backgrounds and rain layers. In this study, we introduce an iterative framework for image deraining to progressively enhance rainy images using a residual long short-term memory structure. The overall network comprises a multidomain residue channel, a fusion module, and a consecutive residual long short-term memory. We introduce multidomain residue channels by computing them in both the image and wavelet lowfrequency domains. We propose a fusion module to combine the residue channel for guidance with wavelet domain features for rain removal. We also propose a feature extraction module based on successive residual long short-term memory to extract the main features in the wavelet domain. An iterative image restoration framework comprising three primary modules is introduced to progressively enhance rainy images. To evaluate the performance of the proposed approach, we conduct experiments using widely used benchmarks. The results demonstrate that our method outperforms state-of-the-art methods in image rain removal https://github.com/ workofsu/.

Images captured under low-light conditions often have various issues, such as low brightness and contrast, and also color distortion. The aim of low-light image enhancement is to improve the visual quality of such images to facilitate subsequent image processing and computer vision tasks. However, this task faces major challenges from the inherent limitations of the low-light environment. This paper proposes a novel low-light image enhancement network using a U-shaped enhanced lightening back-projection. The network architecture primarily comprises a series of lightening blocks and a color balance block. We construct the lightening block, which incorporates a Ushaped enhanced lightening back-projection module and a saturation-guided fusion module, by assuming that the saturation levels of the low-light and the normal images are similar. Additionally, a color balance block is included to address issues such as color fading or over-enhancement. The effectiveness of the proposed network is assessed via extensive experiments conducted on widely-used benchmarks. The simulation results confirm that the proposed low-light enhancement network outperforms other state-of-the-art approaches. https://github. com/JYP-1317/.

The western conifer seed bug (WCSB), Leptoglossus occidentalis, has expanded rapidly in the southern peninsula of Korea since it was first reported in southeastern Korea in 2010. Two types of human-mediated passive movements were devised for modeling the rapid advancement of the pest population in this study: traffic effects and forest-product transportation. A lattice structure model (LSM) was developed to accommodate the traffic effects pertaining to the local area along with the natural population dynamics of the pest. Separately, a spatial network model (SNM) was constructed to present the passive movement of the WCSB because of forest-product transportation between all local areas in Korea. The gravity rule was applied to obtain the parameters for forest-product transportation between the local areas. LSM and SNM were linked to the two present types of passive movements in the model. The model simulated fast, linear advancement in a short period, compared with slow, circular advancement because of the conventional natural diffusion process of populations. Simulation results were comparable to field data observed in the southern peninsula of Korea, matching the rapid advancement of about 400 km to the north area (Seoul) from the south area (Changwon) within six years and expanding across the nation in 10 years. Possible saturation of populations was predicted in the 2020s if survival conditions for the WCSB were favorable and no control efforts were given in field conditions. Dispersal because of SNM notably surpassed the dispersal simulated by LSM when the WCSB population rapidly dispersed over a wide area. The Allee-effect and contribution ratio of SNM were the factors governing the rapid expansion of pest populations. The possibility of using the combined model was further discussed to address different types of human-mediated passive movements associated with population dynamics in forest pest dispersal.

Image and video manipulation has been actively used in recent years with the development of multimedia editing technologies. However, object-based video tampering, which adds or removes objects within a video frame, is posing challenges because it is difficult to verify the authenticity of videos. In this paper, we present a novel object-based frame identification network. The proposed method uses symmetrically overlapped motion residuals to enhance the discernment of video frames. Since the proposed motion residual features are generated on the basis of overlapped temporal windows, temporal variations in the video sequence can be exploited in the deep neural network. In addition, this paper introduces an asymmetric network structure for training and testing a single basic convolutional neural network. In the training process, two networks with an identical structure are used, each of which has a different input pair. In the testing step, two types of testing methods corresponding to two- and three-class frame identifications are proposed. We compare the identification accuracy of the proposed method with that of the existing methods. The experimental results demonstrate that the proposed method generates reasonable identification results for both two- and three-class forged frame identifications.

In sand-dust environments, the low quality of images captured outdoors adversely affects many remote-based image processing and computer vision systems, because of severe color casts, low contrast, and poor visibility of sand-dust images. In such cases, conventional color correction methods do not guarantee appropriate performance in outdoor computer vision applications. In this paper, we present a novel color correction and dehazing algorithm for sand-dust image enhancement. First, we propose an effective color correction method that preserves the consistency of the chromatic variances and maintains the coincidence of the chromatic means. Next, a transmission map for image dehazing is estimated using the gamma correction for the enhancement of color-corrected sand-dust images. Finally, a cross-correlation-based chromatic histogram shift algorithm is proposed to reduce the reddish artifacts in the enhanced images. We performed extensive experiments for various sand-dust images and compared the performance of the proposed method to that of several existing state-of-the-art enhancement methods. The simulation results indicated that the proposed enhancement scheme outperforms the existing approaches in terms of both subjective and objective qualities.

This paper introduces a machine learning based copy-move forgery (CMF) localization method. The basic convolutional neural network cannot be applied to CMF detection because CMF frequently involves rotation transformation. Therefore, we propose a rotation-invariant feature based on the root-mean squared energy using high-frequency wavelet coefficients. Instead of using three color image channels, two-scale energy features and low-frequency subband image are fed into the conventional VGG16 network. A correlation module is used by employing small feature patches generated by the VGG16 network to obtain the possible copied and moved patch pairs. The all-to-all similarity score is computed using the correlation module. To generate the final binary localization map, a simplified mask decoder module is introduced, which is composed of two simple bilinear upsampling and two batch-normalized-inception-based mask deconvolution followed by bilinear upsampling. We perform experiments on four test datasets and compare the proposed method with state-of-the-art tampering localization methods. The results demonstrate that the proposed scheme outperforms the existing approaches.

Images captured under low-light conditions are characterized by problems, including low brightness, low contrast, and color distortion. Low-light image enhancement, aims to enhance visual effects for the benefit of subsequent image processing and computer vision tasks, but it presents a challenging problem. This paper proposes an efficient and fast algorithm for low-light image enhancement. First, it derives a mathematical connection between the Retinex theory and the atmospheric scattering model by normalizing an inverted image to atmospheric light. Next, medium transmission is derived as a function of the saturation of the scene radiance only, and the image-adaptive saturation of scene radiance is estimated via a simple stretch function according to the average saturation of the low-light image. The proposed algorithm is fast because it requires no training, prior knowledge, or refinement. The simulation results confirm that in terms of both computational complexity and enhancement efficiency, the proposed low light enhancement scheme outperforms other state-of-the-art approaches. (c) 2022 Elsevier B.V. All rights reserved.

채도 기반 안개 제거 방법은 속도가 빠르고, 화소 단위로 전달량을 결정하기 때문에 우수한 방법으로 알려져 있다. 그러나, 이 방법은 고정된 채도 스트레칭 방법을 사용하기 때문에, 다양한 채도를 가지는 안개 영상을 적응적으로 제거하지 못하는 단점을 가진다. 본 논문은 적응적 채도 스트레칭 함수를 이용한 효과적인 안개 제거 방법을 제안한다. 먼저 전달량을 구하기 위하여 최대값을 이용하는 HSV 색 공간의 채도를 사용한다. 원 영상의 채도는 안개 영상의 채도에 따른 적응적인 채도 확장 함수를 이용하여 추정한다. 제안 방법은 안개의 양에 따라 채도를 적응적으로 추정할 수 있는 장점을 가진다. 실험 결과를 통하여 제안 방법이 기존의 방법과 비교하여 주관적 및 객관적으로 우수한 성능을 나타냄을 보였다.

Almost all image sensors measure only one color per pixel through the color filter array. Missing pixels are estimated using a demosaicing process. For this reason, a demosaiced image leaves a particular trace. When an image is manipulated or tampered, the demosaicing trace can be changed. This change can serve as a basic clue for detecting or localizing image tampering. Demosaicing pattern-based tampering localization algorithms require a re-interpolation process, and the prediction residue between the given image and the re-interpolated image is commonly used to localize tampered regions. However, the prediction residue is not always valid because the demosaicing interpolation kernel cannot be known, which deteriorates the localization performance. This paper presents an effective re-interpolation process using singular value decomposition for an unknown demosaicing method. First, the green channel of the given image is decomposed into four sub-images according to the Bayer pattern. For a small block of each sub-image, the singular value decomposition is performed. The prediction residue is obtained by reconstructing the image block after removing the largest singular value. The feature to localize the forged regions is extracted by the logarithm ratio of the prediction residue variance. The proposed method does not require any statistical model for the extracted feature, because the prediction residue is more accurate than that of conventional methods. We perform intensive experiments for three test datasets and compare the proposed method with state-of-the-art tampering localization methods, the results of which indicate that the proposed scheme outperforms existing approaches.

Outdoor images in sand-dust environments play an adverse role in various remote-based computer vision tasks because captured sand-dust images have severe color casts, low contrast, and poor visibility. However, although sand-dust image restoration is as important as haze removal and underwater image enhancement, it has not been sufficiently studied. In this paper, we present a novel color balance algorithm for sand-dust image enhancement. The aim of the proposed enhancement method is to obtain a coincident chromatic histogram. First, we introduce a pixel-adaptive color correction method using the mean and standard deviation of chromatic histograms. Pixels of each color component are adjusted based on the statistical characteristics of the green component. Second, a green-mean-preserving color normalization technique is presented. However, using the mean of red and blue components as the mean of the green can result in an undesirable output because the red or blue components of many sand-dust images have a narrow histogram with a high peak. To address this problem, we propose a histogram shifting algorithm that makes the red and blue histograms overlap the green histogram as much as possible. Based on this algorithm, bluish or reddish artifacts of the enhanced image can be reduced. Finally, image adjustment is exploited to improve the brightness of the sand-dust image. We performed intensive experiments for various sand-dust images and compared the performance of the proposed method with those of state-of-the-art enhancement methods. The simulation results indicate that the proposed enhancement scheme outperforms the existing approaches in terms of both subjective and objective qualities.

This Letter presents a simple and new colour correction algorithm for the underwater image. The proposed method corrects the pixels of the weak colour channel based on the ratio of the standard deviation of the weak colour channel to that of the green channel. The corrected image based on the proposed colour balanced algorithm barely produces a reddish artefact and can be applied directly to the existing dehazing approaches without any processing. The experimental results show that the proposed algorithm achieves good colour correction performance, as well as reasonable underwater image enhancement when employed with the existing dehazing methods.

Because digitized images are easily replicated or manipulated, copy-move forgery techniques are rendered possible with minimal expertise. Furthermore, it is difficult to verify the authenticity of images. Therefore, numerous efforts have been made to detect copy-move forgeries. In this paper, we present an improved region duplication detection algorithm based on the keypoints. The proposed algorithm utilizes the scale invariant feature transform (SIFT) and the reduced local binary pattern (LBP) histogram. The LBP values with 256 levels are obtained from the local window centered at the keypoint, which are then reduced to 10 levels. For a keypoint, a 138-dimensional is generated to detect copy-move forgery. We test the proposed algorithm on various image datasets and compare the detection accuracy with those of existing methods. The experimental results demonstrate that the performance of the proposed scheme is superior to that of other tested copy-move forgery detection methods. Furthermore, the proposed method exhibits a uniform detection performance for various types of test datasets.