Low Contrast Areas Research Articles

Deep learning-based segmentation of abdominal aortic aneurysms and intraluminal thrombus in 3D ultrasound images.

Ultrasound (US)-based patient-specific rupture risk analysis of abdominal aortic aneurysms (AAAs) has shown promising results. Input for these models is the patient-specific geometry of the AAA. However, segmentation of the intraluminal thrombus (ILT) remains challenging in US images due to the low ILT-blood contrast. This study aims to improve AAA and ILT segmentation in time-resolved three-dimensional (3D + t) US images using a deep learning approach. In this study a "no new net" (nnU-Net) model was trained on 3D + t US data using either US-based or (co-registered) computed tomography (CT)-based annotations. The optimal training strategy for this low-contrast data was determined for a limited dataset. The merit of augmentation was investigated, as well as the inclusion of low-contrast areas. Segmentation results were validated with CT-based geometries as the ground truth. The model trained on CT-based masks showed the best performance in terms of DICE index, Hausdorff distance, and diameter differences, covering a larger part of the AAA. With a higher accuracy and less manual input the model outperforms conventional methods, with a mean Hausdorff distance of 4.4mm for the vessel and 7.8mm for the lumen. However, visibility of the lumen-ILT interface remains the limiting factor, necessitating improvements in image acquisition to ensure broader patient inclusion and enable rupture risk assessment of AAAs in the future.

A Multi-Step Image Pre-Enhancement Strategy for a Fish Feeding Behavior Analysis Using Efficientnet

To enhance the accuracy of lightweight CNN classification models in analyzing fish feeding behavior, this paper addresses the image quality issues caused by external environmental factors and lighting conditions, such as low contrast and uneven illumination, by proposing a Multi-step Image Pre-enhancement Strategy (MIPS). This strategy includes three critical steps: initially, images undergo a preliminary processing using the Multi-Scale Retinex with Color Restoration (MSRCR) algorithm, effectively reducing the impact of water surface reflections and enhancing the visual effect of the images; secondly, the Multi-Metric-Driven Contrast Limited Adaptive Histogram Equalization (mdc) technique is applied to further improve image contrast, especially in areas of low contrast, by adjusting the local contrast levels to enhance the clarity of the image details; finally, Unsharp Masking (UM) technology is employed to sharpen the images, emphasizing their edges to increase the clarity of the image details, thereby significantly improving the overall image quality. Experimental results on a custom dataset have confirmed that this pre-enhancement strategy significantly boosts the accuracy of various CNN-based classification models, particularly for lightweight CNN models, and drastically reduces the time required for model training compared to the use of advanced ResNet models. This research provides an effective technical route for improving the accuracy and efficiency of an image-based analysis of fish feeding behavior in complex environments.

A gated cross-domain collaborative network for underwater object detection

Underwater object detection (UOD) plays a significant role in aquaculture and marine environmental protection. Considering the challenges posed by low contrast and low-light conditions in underwater environments, several underwater image enhancement (UIE) methods have been proposed to improve the quality of underwater images. However, only using the enhanced images does not improve the performance of UOD, since it may unavoidably remove or alter critical patterns and details of underwater objects. In contrast, we believe that exploring the complementary information from the two domains is beneficial for UOD. The raw image preserves the natural characteristics of the scene and texture information of the objects, while the enhanced image improves the visibility of underwater objects. Based on this perspective, we propose a Gated Cross-domain Collaborative Network (GCC-Net) to address the challenges of poor visibility and low contrast in underwater environments, which comprises three dedicated components. Firstly, a real-time UIE method is employed to generate enhanced images, which can improve the visibility of objects in low-contrast areas. Secondly, a cross-domain feature interaction module is introduced to facilitate the interaction and mine complementary information between raw and enhanced image features. Thirdly, to prevent the contamination of unreliable generated results, a gated feature fusion module is proposed to adaptively control the fusion ratio of cross-domain information. Our method presents a new UOD paradigm from the perspective of cross-domain information interaction and fusion. Experimental results demonstrate that the proposed GCC-Net achieves state-of-the-art performance on four underwater datasets.

CSwinDoubleU-Net: A double U-shaped network combined with convolution and Swin Transformer for colorectal polyp segmentation

The early detection of polyps from colonoscopy can reduce the risk of colorectal cancer development and give timely treatment. Accurate polyp segmentation during colonoscopy examinations can help clinicians locate the polyps which is of great significance in the clinical prevention of colorectal cancer. However, due to large variations in terms of size, color, texture, and morphology of polyps, the similarity between the polyp lesions and their background, the variations of illumination, motion blur, low-contrast areas, intestinal contents during image acquisitions, precisely polyp segmentation is still an open issue. To overcome the challenges above, a new double U-shaped image segmentation network combining convolution structure and Shifted Windows (Swin) Transformer, namely CSwinDoubleU-Net, is proposed in this paper. The developed CSwinDoubleU-Net consists of two U-shaped encode-decode structures associated with a pure CNN-based structure and a pure Transformer-based one, and a feature fusion module that interacts with the two U-shaped parts. The first U-shaped codecs network utilizes the multi-convolution layers to extract the local feature information and the coordinate attention module in each skip connection to effectively reduce the loss of spatial information and obtain the accurate position information of the encoded features. Next, the second U-shaped codecs network uses the Swin Transformer layers with the sliding window to extract global feature information further. At last, a convolutional feature and self-attention feature fusion module (CSFFM) is designed to deeply fuse the local convolutional features extracted from the first U-shape structure and the global self-Attention features extracted from the second U-shape structure. The obtained multi-category and multi-dimensional fused feature information can help to recover the boundary features of polyps. Extensive experiments are conducted to validate the proposed CSwinDoubleU-Net on five publicly available datasets, including CVC-ClinicDB, Kvasir, CVC-ColonDB, CVC-T, and ETIS-Larib. The results show that the proposed model can outperform some state-of-the-art methods and achieve high segmentation performance for polyp images.

Thresholding-accelerated convolutional neural network for aeroengine turbine blade segmentation

Turbine blades can only be detected nondestructively and precisely using industrial computed tomography (CT). The accuracy of CT image segmentation, which is a key step in CT detection, significantly affects CT detection precision. However, CT images of turbine blades cannot be segmented precisely because they contain inhomogeneity, noise, artifacts, and low-contrast areas. In addition, it is crucial to consider both efficiency and precision in industrial applications. Unfortunately, large-size networks often tend to be inefficient, while lightweight networks lack the necessary accuracy. To address these issues, we propose an innovative approach utilizing an ultra-lightweight neural network that fuses a threshold-based segmentation algorithm with a convolutional neural network for acceleration. The integrated thresholding algorithm was developed through the convolution of the Phansakar method. The convolutional neural network was designed with a simplified multichannel convolution operation to minimize computational cost. Notably, the proposed neural network utilizes only 0.11 M learnable parameters, which is significantly less than the number utilized by most classical deep neural networks. Experimental results for CFM56-7BE and Pratt & Whitney F100 aeroengine hollow turbine blade CT images demonstrate that compared to state-of-the-art algorithms, our proposed network can achieve 0.44 % mean intersection over union and 0.23 % mean Dive similarity coefficient improvements on testing datasets with significantly reduced computational cost. Therefore, the proposed method has high practical application significance and can achieve real-time detection.

Simulation of daylight availability, visual comfort and view clarity for a novel window system with switchable blinds in classrooms

The presence of daylight reflections on blackboards in classrooms can be disruptive to students and negatively impact their visual comfort and learning efficiency. A novel window system with switchable blinds has been specifically designed to balance daylight availability, visual comfort, and adequate window view in classrooms. This study aims to develop a validated simulation method to assess the daylighting performance of the novel window system in these three domains. Radiance-based simulation using bidirectional scattering distribution function (BSDF) data was developed and validated against experimental measurements in a test room. Daylight autonomy (DA) was used as the criterion for daylight availability, while the contrast ratio on the board was used to evaluate the disturbing light reflections on it. The simulation results show that, regardless of the classroom orientation (south/east/west), the novel window system with switchable blinds can: a) successfully meet the LEED requirements for spatial daylight autonomy (sDA) > 55%; b) reduce the low-contrast areas on the blackboard by adaptively switching the blinds, and c) maintain a satisfactory openness of the window with a clear image of the exterior. The studied window system with switchable blinds can provide a sensible balance between daylight availability, visual comfort, and view. The method of rating visual comfort in classrooms using contrast ratio on the blackboard was demonstrated to be feasible and applicable.

CAMS-Net: An attention-guided feature selection network for rib segmentation in chest X-rays

Segmentation of clavicles and ribs in chest X-rays is significant for diagnosing lung diseases. However, it is challenging to segment ribs because of the low contrast on chest X-rays. Moreover, most existing methods fail to segment ribs in the area of abnormal gray value caused by overlapping anatomical structures or lesions. A novel algorithm based on attention-guided feature selection named CAMS-Net is presented in this paper, aiming to improve the accuracy of ribs segmentation in low contrast areas and abnormal gray value areas. The collaborative attention skip connection module (CAS) introduces the decoder features and attention-guided feature selection into the traditional skip connection, which highlights the feature representation of ribs. The attention-guided multi-scale feature selection module (AMFS) increases receptive field size to connect the abnormal rib gray value region with the normal gray value region. To reduce the influence of background pixels, the AMFS selects important features through attention and uses multi-scale information to jointly decide the final class of the pixel. The paper conducted extensive experiments to evaluate CAMS-Net. Compared with state-of-the-art methods, the average values of Precision and Jaccard are increased by 0.89% and 1.23%, respectively. The Recall and Jaccard values of the anterior rib are increased by 2.64% and 2.52%, respectively. Qualitative analysis shows that CAMS-Net can improve the segmentation accuracy of low-contrast areas and maintain the segmentation integrity in areas with abnormal gray values. The robustness and generalization of CAMS-Net are verified through external tests on VinDr-RibCXR, JSRT, Shenzhen, and NIH datasets. Besides, CAS and AMFS modules can be flexibly inserted into other backbone networks.

Dynamic Range Compression of Thermograms for Assessment of Welded Joint Face Quality.

Temperature is one of the essential parameters in fusion welding. Typically, an uncooled infrared detector acquires 14-bit data, while a human observer can only distinguish about 128 levels of grey. For IR HDR (high dynamic range) images, one of the main goals of dynamic range compression is to enhance the visibility of low-contrast details. It is an important issue because the temperature span in the cross-section of a welded joint and its length are large. In the paper, global approaches for range compression are investigated, such as algorithms that include pixel transformations, histogram equalization ('he') and some of its variants. Additionally, multiscale decomposition methods were investigated. All results are obtained for the sequences of thermograms acquired during the TIG welding of plates made of Inconel 625 superalloy. The process was observed with an uncooled IR camera. The application of compression methods led to the generation of low-dynamic-range (LDR) IR images. The algorithms allowed the preservation of global contrast and enhancement of the visibility of hot details in dark and low-contrast areas. All IR representations of the welded samples were evaluated, and relationships between apparent temperature counted in the pixel-level value and weld-face geometry were revealed. Methods based on wavelet transforms were found to be the most suitable for this type of image; nevertheless, a relatively large local noise was generated.

A Robust Multiscale Edge Detection Method for Accurate SAR Image Registration

Edge detection is a technique used to identify inherent structures within an image, and it is an essential requirement for synthetic aperture radar (SAR) applications. In particular, ratio-based edge detectors have been widely used in SAR image registration because of their ability to extract invariant features and reduce the effects of speckle noise. However, current edge detectors often struggle to accurately detect multi-scale objects and low-contrast structures. To address this issue, we present a robust multi-scale edge detector that uses a modified convolution kernel to improve the extensibility of edge features and aggregates multi-scale feature responses. We also propose a local scale estimation module to enhance edge responses in low-contrast areas and reduce noise effects. The experimental results demonstrate that our proposed method effectively preserves the integrity, continuity, and robustness of multi-scale and low-contrast structures. By incorporating our proposed edge detector into feature and template matching frameworks, we are able to significantly improve matching accuracy and outperform state-of-the-art SAR image registration methods.

Deep learning enhanced NIR-II volumetric imaging of whole mice vasculature

Fluorescence imaging through the second near-infrared window (NIR-II,1000–1700 nm) allows in-depth imaging. However, current imaging systems use wide-field illumination and can only provide low-contrast 2D information, without depth resolution. Here, we systematically apply a light-sheet illumination, a time-gated detection, and a deep-learning algorithm to yield high-contrast high-resolution volumetric images. To achieve a large FoV (field of view) and minimize the scattering effect, we generate a light sheet as thin as 100.5 μm with a Rayleigh length of 8 mm to yield an axial resolution of 220 µm. To further suppress the background, we time-gate to only detect long lifetime luminescence achieving a high contrast of up to 0.45 <italic>Ι</italic>_contrast. To enhance the resolution, we develop an algorithm based on profile protrusions detection and a deep neural network and distinguish vasculature from a low-contrast area of 0.07 <italic>Ι</italic>_contrast to resolve the 100 μm small vessels. The system can rapidly scan a volume of view of 75 × 55 × 20 mm³ and collect 750 images within 6 mins. By adding a scattering-based modality to acquire the 3D surface profile of the mice skin, we reveal the whole volumetric vasculature network with clear depth resolution within more than 1 mm from the skin. High-contrast large-scale 3D animal imaging helps us expand a new dimension in NIR-II imaging.

Deep learning-based noise reduction for coronary CT angiography: using four-dimensional noise-reduction images as the ground truth.

To assess low-contrast areas such as plaque and coronary artery stenosis, coronary computed tomography angiography (CCTA) needs to provide images with lower noise without increasing radiation doses. To develop a deep learning-based noise-reduction method for CCTA using four-dimensional noise reduction (4DNR) as the ground truth for supervised learning. \We retrospectively collected 100 retrospective ECG-gated CCTAs. We created 4DNR images using non-rigid registration and weighted averaging three timeline CCTA volumetric data with intervals of 50 ms in the mid-diastolic phase. Our method set the original reconstructed image as the input and the 4DNR as the target image and obtained the noise-reduced image via residual learning. We evaluated the objective image quality of the original and deep learning-based noise-reduction (DLNR) images based on the image noise of the aorta and the contrast-to-noise ratio (CNR) of the coronary arteries. Further, a board-certified radiologist evaluated the blurring of several heart structures using a 5-point Likert scale subjectively and assigned a coronary artery disease reporting and data system (CAD-RADS) category independently. DLNR CCTAs showed 64.5% lower image noise (P < 0.001) and achieved a 2.9 times higher CNR of coronary arteries than that in original images, without significant blurring in subjective comparison (P > 0.1). The intra-observer agreement of CAD-RADS in the DLNR image was excellent (0.87, 95% confidence interval = 0.77-0.99) with original CCTAs. Our DLNR method supervised by 4DNR significantly reduced the image noise of CCTAs without affecting the assessment of coronary stenosis.

Contrast Enhancement Method Based on Multi-Scale Retinex and Adaptive Gamma Correction

The most common methods to improve the quality of images with insufficient visibility are retinex-based and gamma correction methods. The fundamental assumption of retinex theory is that the color of an object can be represented as the multiplication of its illumination and reflectance. The retinex-based method improves the quality of the insufficiently visible image by repairing its illumination. The multi-scale retinex (MSR) is a classic retinex-based method. Though MSR better enhances the details of the image, it sometimes reverses its lightness value. The method based on adaptive gamma correction with weighting distribution (AGCWD) is to modify the visibility of images by gamma function. However, AGCWD provides a good enhancement effect on low-contrast areas, it also enhances the high-light region making it too bright. In this paper, a method that combines the advantages of MSR and AGCWD is proposed. Firstly, the advatages of MSR and AGCWD are preserved into detailed image through the weight that considers illumination. Then, the image constructed by combining the detailed and original images could maintain the contrast of the high-light region and enhance details of the low-light region. The validity of the proposed method is shown by experiments using several images.

The impact of ASiR-V on abdominal CT radiation dose and image quality – A phantom study

IntroductionTo investigate how ASiR-V and kVp changes in Computed tomography (CT) affect radiation dose and image quality, when using automatic tube current modulation (ATCM) for different sized phantoms. MethodsA liver-phantom with two different liver inserts (QRM, Moehrendorf, Germany), with extension rings, representing fat, were additionally applied to the phantom to simulate patients of different sizes (small: 30cm diameter, medium: 35cm and large: 40cm). Abdominal scans were performed on a 256 slice CT scanner (GE Healthcare, Milwaukee, WI, USA), with consistent pitch (0.992), rotation time (0.5s), slice thickness (0.625mm) and collimation (80mm), while other parameters were varied (kVp: 80/100/120/140; Noise Index: 13/22; mA interval 80-720, ASiR-V: 30/60/100%). CTDI and DLP was recorded for each scan and image quality was assessed using objective metrics in predefined anatomic areas (HU and noise). Radiation dose and image quality metrics were compared between protocols. ResultsCTDI decreased by 80% from ASIR-V 30% to ASiR-V 100% for prescribed NI 13, and by 79% for the prescribed NI of 22. For 100% ASiR-V and a prescribed NI of 22 the CTDI remained the same regardless of phantom size for the different kVp settings. Pairwise comparison revealed significant differences in CTDI (p < 0.0001) for all combinations of prescribed NI and ASIR-V levels, except the difference between ASIR-V levels of 30 and 60%, with a prescribed NI of 13 (p = 0.124).When data from the three phantom sizes were combined, increasing ASIR-V from 30-100%, resulted in noise decreases of 22% for NI of 13 and by 8% for NI of 22. Notably, image quality in the low contrast area of the liver insert was impaired when the large phantom was scanned with 100% ASiR-V and either 80/100kVp (NI 22), because of the large reduction in tube current applied (down to 80 mA). ConclusionSubstantial radiation dose reductions (up to 80%) resulted from increasing ASiR-V levels. However, image quality deteriorates when 100% ASiR-V is applied due to low applied tube current by the ATCM.

IMPROVED EDGE DETECTION FOR SATELLITE IMAGES

Abstract. Edges are a key feature employed in various computer vision applications namely segmentation, object recognition, feature tracking and 3D reconstruction. Edges provide key information with regards to object presence, shape, form and detail which aid in many computer vision tasks. While there are various edge detection techniques in literature, challenges in edge detection remain. Varying image contrast due to non uniform scene illumination and imaging resolution affects the edge information obtained from any given image. The edge detection results are characterised by missing edges, edge fragmentation and some false positive edges. Gradient based edge detectors are the most commonly used detectors. These detectors all suffer from aforementioned challenges. In this, paper we present an edge detection framework that aims to recover long unfragmented edges from satellite images. This is achieved by using an edge accumulator that operates on the entire edge detection parameter space. Gradient based edge detectors rely on thresholding to retrieve salient edges. This usually results in missed or noisy edges. To counter this, the accumulator is run over a wide parameter space, growing edges at each accumulator level while maintaining edge position using a localisation filter. The results are longer unbroken edges that are detected for most objects, even in shadowy regions and low contrast areas. The results show improved edge detection that preserves the form and detail of objects when compared to current gradient based detectors.

Intraoral 3-D Measurement by Means of Group Coding Combined With Consistent Enhancement for Fringe Projection Pattern

Intraoral 3D measurement can obtain digital impressions in real time. However, owing to saliva, enamel, metallic denture, etc., the quality of the captured 2D image using the intraoral equipment, which is used for feature measurement and 3D reconstruction, is usually degraded. In this paper, we propose a equipment with high performance for intraoral 3D measurement. First, group-multi-line coding algorithm instead of sinusoidal or gray code for fringe projection profilometry (FPP) was applied for stripe pattern decoding. The distance between adjacent stripes was large enough to avoid the scatter of the tooth. It can also overcome the shortcoming of sinusoidal-based phase computing, leading to accuracy degradation owing to scanner jittering. Second, a consistent stripe feature enhancement method is proposed. The stripe feature in the mirror, exposure, and low-contrast area was homogeneously enhanced. Consequently, an individual scene is reconstructed using more effective and accurate points. This not only improves the precision of point cloud registration, it also reduces the probability of mismatch and shortens the scanning time for a full-arch. Our results showed that the accuracy of intraoral 3D measurements conducted through the proposed equipment by using novel algorithms increased to 10 - 20 &#x03BC;m and the measurement efficiency increased to 30%.

Enhanced Security: Implementation of Hybrid Image Steganography Technique using Low-Contrast LSB and AES-CBC Cryptography

Now-a-days, sensitive and confidential information needs to be exchanged over open, public, and not secure networks such as the Internet. For this purpose, some information security techniques combine cryptographic and steganographic algorithms and image processing techniques to exchange information securely. Therefore, this research presents the implementation of an algorithm that combines the AES-CBC cryptographic technique with the LSB steganographic technique, which is statistically enhanced by image processing by looking for low-contrast areas where the encrypted information will be stored. This hybrid algorithm was developed to send a plaintext file hidden in an image in BMP format, so the changes in the image are invisible to the human eye and undetectable in possible steganographic analysis. The implementation was performed using Python and its libraries PyCryptodome for encryption and CV2 for image processing. As a result, it was found that the hybrid algorithm implemented has three layers of security over a plaintext encrypted and hidden in a digital image, which makes it difficult to break the secrecy of the information exchanged in a stego-image file. Additionally, the execution times of the hybrid algorithm were evaluated for different sizes of plaintext and digital image files.

Joint Nonlocal, Spectral, and Similarity Low-Rank Priors for Hyperspectral–Multispectral Image Fusion

The fusion of a low-spatial-and-high-spectral resolution hyperspectral image (HSI) with a high-spatial-and-low-spectral resolution multispectral image (MSI) allows synthesizing a high-resolution image (HRI), supporting remote sensing applications such as disaster management, material identification, and precision agriculture. Unlike existing variational methods using low-rank regularizations separately, we present an HSI-MSI fusion method promoting various low-rank regularizations jointly. Our method refines the HRI spatial and spectral correlations from the individual HSI and MSI data through the proper plug-and-play (PnP) of a nonlocal patch-based denoiser in the alternating direction method of multipliers (ADMM). Notably, we consider the nonlocal self-similarity, the spectral low-rank, and introduce a rank-one similarity prior. Furthermore, we demonstrate via an extensive empirical study that the rank-one similarity prior is an inherent characteristic of the HRI. Simulations over standard benchmark datasets show the effectiveness of the proposed HSI-MSI fusion outperforming state-of-the-art methods, particularly in recovering low-contrast areas.

A Neural Network-Based Algorithm for Weak Signal Enhancement in Low Illumination Images

There is noise interference in low-illumination images, which makes it difficult to extract weak signals. For this reason, this paper proposes a low-illumination image weak signal enhancement algorithm based on neural network. Multi-scale normalization is performed on low-light images, and multi-scale Retinex is used to enhance weak signals in low-light images. On this basis, the GAN artificial neural network is used to detect the weak signal of the weak signal in the image, the normalization of the weak signal of the low-illumination image is completed based on the residual network, the self-encoding parameters of the depth residual are generated, and the weak signal enhancement result of the low-illumination image is output. The experimental results show that the method in this paper has better enhancement effect on low-illumination images and better image denoising effect. When the scale value is large, the low-contrast area of the low-illumination image has a better enhancement effect. The saturated area of the low-light image has a better enhancement effect.

Association Between the Severity of Diabetic Retinopathy and Optical Coherence Tomography Angiography Metrics.

Diabetic retinopathy, the most serious ocular complication of diabetes, imposes a serious economic burden on society. Automatic and objective assessment of vessel changes can effectively manage diabetic retinopathy and prevent blindness. Optical coherence tomography angiography (OCTA) metrics have been confirmed to be used to assess vessel changes. The accuracy and reliability of OCTA metrics are restricted by vessel segmentation methods. In this study, a multi-branch retinal vessel segmentation method is proposed, which is comparable to the segmentation results obtained from the manual segmentation, effectively extracting vessels in low contrast areas and improving the integrity of the extracted vessels. OCTA metrics based on the proposed segmentation method were validated to be reliable for further analysis of the relationship between OCTA metrics and diabetes and the severity of diabetic retinopathy. Changes in vessel morphology are influenced by systemic risk factors. However, there is a lack of analysis of the relationship between OCTA metrics and systemic risk factors. We conducted a cross-sectional study that included 362 eyes of 221 diabetic patients and 1,151 eyes of 587 healthy people. Eight systemic risk factors were confirmed to be closely related to diabetes. After controlling these systemic risk factors, significant OCTA metrics (such as vessel complexity index, vessel diameter index, and mean thickness of retinal nerve fiber layer centered in the macular) were found to be related to diabetic retinopathy and severe diabetic retinopathy. This study provides evidence to support the potential value of OCTA metrics as biomarkers of diabetic retinopathy.

Content-Sensitive Superpixel Generation for SAR Images With Edge Penalty and Contraction–Expansion Search Strategy

In this article, we present a content-sensitive superpixel generation method with edge penalty and the contraction–expansion search strategy (EPCES) for synthetic aperture radar (SAR) images. Specifically, the edge information can be obtained by our previously proposed ratio-based edge detector with recurrent guidance filter, which has been proven to be robust to speckle noise and capable of detecting weak edges in low-contrast areas. The content-sensitive superpixel seeds’ initialization method is proposed with respect to the heterogeneous state of the SAR imagery, benefiting from which EPCES can generate an exact number of superpixels set by the user and the fine details can be preserved well. In EPCES, a new dissimilarity with edge penalty is defined to generate the superpixels with better edge adherence. Rather than adopting the conventional clustering method based on local <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$k$ </tex-math></inline-formula> -means, we propose the contraction–expansion search strategy (CES), which explicitly utilizes the continuity information contained in neighboring pixels and enforces the connectivity of the superpixel without any postprocessing step. With the aid of the CES, our proposed method can attain superpixels with low computational cost and high edge adherence. Experimental results on both synthetic and real-world SAR images verify that the proposed method consistently performs favorably against several state-of-the-art methods in terms of both quality and efficiency.