首页期刊导航|The imaging science journal
期刊信息/Journal information
The imaging science journal
Royal Photographic Society of Great Britain
The imaging science journal

Royal Photographic Society of Great Britain

季刊

1368-2199

The imaging science journal/Journal The imaging science journal
正式出版
收录年代

    Video enhancement for dense haze removal using an optimized multi-task evolutionary artificial neural network

    Vishalkirthi S. PatilRaviraj H. Havaldar
    267-282页
    查看更多>>摘要：Images captured during cloudy or misty weather often suffer from poor contrast, colour distortions, and limited visibility. To overcome these challenges, this manuscript proposes a Video Enhancement for Dense Haze Removal using an Optimized Multi-Task Evolutionary Artificial Neural Network (VE-MTEANN-BWOA-DHR). Initially, input videos are sourced from the Real Haze Video Database. Then Adaptive Self-Guided Filtering (ASGF) for eliminate noise from the input video. Then Ternary Pattern with Discrete Wavelet Transform (TPDWT) is used to extract the features. The extracted features are given to a Multi-Task Evolutionary Artificial Neural Network (MTEANN) to classify the dense haze levels in video frames as hazy image 1, hazy image 2, hazy image 3, and hazy image 4. Typically, MTEANN lacks adaptive optimization strategies to determine ideal parameters for effective video enhancement. Therefore, the Beluga Whale Optimization Algorithm (BWOA) is utilized to optimize MTEANN. The proposed VE-MTEANN-BWOA-DHR method demonstrates superior performance compared to the existing models.
      原文链接:
    • NETL
    • NSTL
    • Taylor & Francis

    Face photo-drawing conversion based on multi-scale feature-enhanced generative adversarial networks

    Po WangYi Lihamu Ya Ermaimaiti
    283-298页
    查看更多>>摘要：This paper introduces a novel face photo-to-sketch synthesis method using a multi-scale feature-enhanced generative adversarial network (MFEGAN). The MFEGAN framework captures features at various scales through a multi-scale feature extraction module, enhanced by an attention mechanism. An improved attention residual block in the generator adaptively refines deep image features, improving overall quality. A pre-trained feature extraction network extracts and fuses face-specific features, enriching identity information. Multi-scale perceptual and focal frequency losses optimize detail quality, aligning with human perception. Experimental results show that MFEGAN outperforms existing methods in visual appeal and fidelity to original identity features.
      原文链接:
    • NETL
    • NSTL
    • Taylor & Francis

    Minimum error threshold segmentation method for SAR image based on Rayleigh distribution assumption

    Yang Lv
    299-309页
    查看更多>>摘要：Accurately extracting and segmenting boundaries between targets and backgrounds is difficult in Synthetic Aperture Radar (SAR) images because of the many scattering mechanisms that produce echo signals. These complications combined with the ubiquitous speckle noise present in SAR images cause traditional segmentation algorithms to frequently produce unacceptable results. In order to tackle these problems, we provide a unique segmentation strategy that makes use of a minimum error thresholding technique based on the theory of the Rayleigh distribution. Our technique builds a local translation Rayleigh model by utilizing the stationary wavelet domain, which successfully suppresses speckle noise and generates SAR images that are easier to segment. Next, we utilize a two-dimensional entropy thresholding method to segment images. We present a variable code length genetic technique that incorporates the optimization of segmentation category numbers encoded in the chromosomes into the fitness function in order to further improve segmentation performance. Furthermore, our method yields minimum segmentation error pixels and excellent segmentation efficiency since the segmentation thresholds it finds is accurate. By addressing major issues with speckle noise and segmentation precision, this enhanced approach provides notable improvements in the precise and effective segmentation of SAR images.
      原文链接:
    • NETL
    • NSTL
    • Taylor & Francis

    Squirrel Search Optimization-based near-duplicate image detection

    Srinidhi SundaramKamalakkannan SomasundaramSasikala Jayaraman
    310-327页
    查看更多>>摘要：Near duplicate (ND) image detection is a significant issue in a modern online environment with a wide range of applications like the detection of copyright violations and saving of storage space. Several existing ND detection techniques are perhaps not suitable for online applications due to the large computational burden, and may not successfully detect NDs containing large smooth and plain regions. In addition, the K-means algorithm used in most of the existing methods yield sub-optimal quantization of visual words. This article employs a robust algorithm of Squirrel Search Optimization (SSO) for quantization, fast-hessian matrix-based detector (FHMBD) and FAST Corner Detector (FCD) for the detection of KPs at both plain and non-smooth regions of all images, SURF for computing descriptors and Principal Component Analysis (PCA) for dimensionality reduction. The results of the developed method presented on five image databases. The proposed method offered 99.9% accuracy, 98.67% sensitivity, and 99.91% specificity respectively.
      原文链接:
    • NETL
    • NSTL
    • Taylor & Francis

    Advancements in adversarial generative text-to-image models: a review

    Rawan ZaghloulEnas RawashdehTomader Bani-Ata
    328-353页
    查看更多>>摘要：This comprehensive study explores the landscape of Text-to-image Generative Adversarial Networks (T2I-GANs), aiming to spot the light on their architecture, evaluation methodologies, and limitations. It begins with an overview of the GAN paradigm and then classifies and analyzes various models based on their architecture. The study examines the most common evaluation metrics and datasets used in the field, providing detailed comparisons that offer insight into models' architectures and performances. Additionally, it discusses the diverse experiments performed for model assessment and the limitations reported in existing research, highlighting challenges and potential areas for improvement. This exploration aims to serve as a valuable resource for researchers, practitioners, and enthusiasts interested in the evolving domain of T2l-GANs.
      原文链接:
    • NETL
    • NSTL
    • Taylor & Francis

    Unsupervised low-light image enhancement by data augmentation and contrastive learning

    Shao JunzheZhang Zhibin
    354-362页
    查看更多>>摘要：Today, with the increasing demand for visual perception and high-level computational vision tasks, the field of low-light enhancement is rapidly developing. However, models trained on existing datasets often fail or suffer significant performance degradation in real-world low-light scenarios. This performance degradation is frequently due to the limitations of current databases, which typically contain small quantities of paired images of a single type. This article proposes an unsupervised model with a unique data augmentation technique that transforms a regular image database into a paired image database. By adjusting image parameters during training to change exposure, a regular image database can be converted into a paired one. The model restores low-exposure images by extracting lighting features through comparative learning. Evaluations of the LOL and DIV2K datasets demonstrate the proposed model's effectiveness, achieving notable results in low-light image enhancement. This method removes dataset restrictions, broadening the model's range of applications.
      原文链接:
    • NETL
    • NSTL
    • Taylor & Francis

    Enhancing image encryption security through integration multi-chaotic systems and mixed pixel-bit level

    Muhammad Naufal Erza FarandiAris MarjuniNova RijatiDe Rosal Ignatius Moses Setiadi...
    363-380页
    查看更多>>摘要：This research proposes a new image encryption method to improve the security of image encryption by integrating a multi-chaotic system and mixed pixel-bit level encryption. In the face of the growing use of communications and computer technology, as well as the need to protect sensitive information, this method utilizes chaotic algorithms to increase encryption security. Experimental results show increased security and resistance to various attacks, validating the method's effectiveness in protecting digital images. The measurement tools used include histograms, information entropy, adjacent pixel correlation coefficient, key sensitivity analysis, and robustness testing. In conclusion, this method offers an effective and robust approach to image encryption, providing an important contribution to the field of information security.
      原文链接:
    • NETL
    • NSTL
    • Taylor & Francis

    Efficient and robust techniques for infrared imaging system correction

    Sid Ahmed HamadoucheAyoub BoutemedjetAzzedine Bouaraba
    381-400页
    查看更多>>摘要：Stripe noise is a prevalent issue in infrared imaging systems, characterized by its distinctive directional features, which often appear as vertical lines across the image. This type of noise can significantly degrade the quality of the captured images, making it crucial to address and mitigate its effects. This paper presents an effective strategy to tackle this problem by transforming it from a 2D image issue into a 1D signal problem, enabling efficient resolution of stripes in infrared images. By understanding the characteristics of stripe noise, the proposed algorithm effectively solves the problem by first computing the column average of the noisy image, extracting stripe components from this one-dimensional signal, and effectively removing the stripes without blurring image details. This approach has been tested on numerous images with varying noise levels, demonstrating exceptional denoising performance compared to state-of-the-art methods. The results show marked improvements in visual quality, especially around edges and smooth areas, without requiring complex algorithms or iterative processes.
      原文链接:
    • NETL
    • NSTL
    • Taylor & Francis

    Intelligent extraction of salient objects from fuzzy distorted images based on multi-scale convolution

    Lingling Lu
    401-414页
    查看更多>>摘要：To solve the problem of how to obtain important information from fuzzy distorted images, an intelligent method for extracting salient objects from fuzzy distorted images based on multi-scale convolution is proposed. A multi-scale deformation feature convolution network is established to extract the salient features of fuzzy distorted images. Among them, the multi-scale convolution network VGG-16 extracts the shallow features of the fuzzy distortion image through the convolution layer operation, uses the multi-scale convolution kernel to extract the fuzzy distortion image in parallel, and introduces the self-attention mechanism to make the extracted features more relevant, Input the fused features into the proposed area extraction network, Through the target area detection network, the proposed target area of the fuzzy distortion image is classified and position regression is performed to obtain the final salient target of the fuzzy distortion image. The experimental results show that this method can accurately recognize the contour of salient objects, and extract salient objects from fuzzy distorted images; The extracted salient objects have high accuracy, low mean square error and similarity of 92.9% with the original image; For the salient target extraction in various scenarios, it shows high advantages.
      原文链接:
    • NETL
    • NSTL
    • Taylor & Francis

    Rectifying inhomogeneous illumination in digital images using a latent light manifestation algorithm

    Zohair Al-Ameen
    415-427页
    查看更多>>摘要：Rectification of illumination is vital in enhancing the visibility of images acquired in suboptimal lighting conditions. Still, it remains a challenging yet intriguing process as many existing algorithms fail to meet expectations. Hence, a latent light manifestation (LLM) algorithm is introduced, beginning with a conversion to the HSV space, processing the V channel using the Naka-Rushton and exponential equations. Next, the two output images are blended using a two-step range-expansion blending approach. After that, a statistics-based tonality adjustment method and a dynamic range extension process are applied. Finally, a transfer to the RGB space is performed to produce the output image. The LLM is tested using three datasets, compared with eight contemporary algorithms, and the quality evaluations are carried out via five performance measures. The results denote that the developed LLM consistently outperforms existing algorithms, scoring the best according to the used measures and yielding visually pleasing results.
      原文链接:
    • NETL
    • NSTL
    • Taylor & Francis