Superpixel Image Research Articles

Satellite IoT Based Road Extraction from VHR Images Through Superpixel-CNN Architecture

In the past few decades, technology has progressively become ineluctable in human lives, primarily due to the growth of certain fields like space technology, Big Data, the Internet of Things (IoT), and machine learning. Space technology has revolutionized communication mechanisms while creating opportunities for various research areas, including remote sensing (RS)-inspired applications. On the other hand, IoT presents a platform to use the power of the internet over a whole range of devices through a phenomenon known as social IoT. These devices generate a humongous amount of data that requires handling and managing by big data technology incorporated with deep learning techniques to reduce the manual workload of an operator. Moreover, deep learning architectures like convolutional neural networks (CNNs) have presented a scope to extract the underlying features from the large-scale input images in providing better solutions for tasks such as automatic road detection that come at the cost of time and memory overhead. In this context, we have proposed a three-layer edge-fog-cloud-based intelligent satellite IoT architecture that uses the superpixel-based CNN approach. At the fog layer, the superpixel-based simple linear iterative cluster (SLIC) algorithm uses the images captured by the satellites of the edge level to produce the smaller-sized superpixel images that can be transferred even in a low bandwidth link. The CNN module at the cloud level is then trained with these superpixel images to predict the road networks from these RS images. Two popular road datasets: the DeepGlobe Road dataset and the Massachusetts Road dataset, have been considered to prove the usefulness of the proposed SLIC-CNN architecture in satellite-based IoT platforms to address the problems like RS image-based road extraction. The proposed architecture achieves better performance accuracy than the classical CNN while reducing the incurred overhead by a noticeable limit. • Presents a superpixel-inspired deep learning-based architecture that optimizes the time and memory overhead. • Presents the three-layer edge-fog-cloud-based satellite IoT architecture to deploy the proposed SLIC-CNN framework. • Discusses the efficiency of the proposed deep architecture to work in a low bandwidth environment for road segmentation. • Highlights the application areas such as emergency alarm systems that can use the architecture to operate in real-time.

Multi Level Approach for Segmentation of Interstitial Lung Disease (ILD) Patterns Classification Based on Superpixel Processing and Fusion of K-Means Clusters: SPFKMC.

During the COVID-19 pandemic, huge interstitial lung disease (ILD) lung images have been captured. It is high time to develop the efficient segmentation techniques utilized to separate the anatomical structures and ILD patterns for disease and infection level identification. The effectiveness of disease classification directly depends on the accuracy of initial stages like preprocessing and segmentation. This paper proposed a hybrid segmentation algorithm designed for ILD images by taking advantage of superpixel and K-means clustering approaches. Segmented superpixel images adapt the better irregular local and spatial neighborhoods that are helpful to improving the performance of K-means clustering-based ILD image segmentation. To overcome the limitations of multiclass belongings, semiadaptive wavelet-based fusion is applied over selected K-means clusters. The performance of the proposed SPFKMC was compared with that of 3-class Fuzzy C-Means clustering (FCM) and K-Means clustering in terms of accuracy, Jaccard similarity index, and Dice similarity coefficient. The SPFKMC algorithm gives an accuracy of 99.28%, DSC 98.72%, and JSI 97.87%. The proposed Fused Clustering gives better results as compared to traditional K-means clustering segmentation with wavelet-based fused cluster results.

Superpixel Image Classification with Graph Convolutional Neural Networks Based on Learnable Positional Embedding

Graph convolutional neural networks (GCNNs) have been successfully applied to a wide range of problems, including low-dimensional Euclidean structural domains representing images, videos, and speech and high-dimensional non-Euclidean domains, such as social networks and chemical molecular structures. However, in computer vision, the existing GCNNs are not provided with positional information to distinguish between graphs of new structures; therefore, the performance of the image classification domain represented by arbitrary graphs is significantly poor. In this work, we introduce how to initialize the positional information through a random walk algorithm and continuously learn the additional position-embedded information of various graph structures represented over the superpixel images we choose for efficiency. We call this method the graph convolutional network with learnable positional embedding applied on images (IMGCN-LPE). We apply IMGCN-LPE to three graph convolutional models (the Chebyshev graph convolutional network, graph convolutional network, and graph attention network) to validate performance on various benchmark image datasets. As a result, although not as impressive as convolutional neural networks, the proposed method outperforms various other conventional convolutional methods and demonstrates its effectiveness among the same tasks in the field of GCNNs.

China's oases have expanded by nearly 40% over the past 20 years

AbstractOases are supporters of biological, social, and economic activities in arid regions, which guides their prosperity or decay. Given that China has the largest oases area in the world, determining their evolution is of substantial importance for sustainable development and the ecological protection of these systems. Accordingly, the spatiotemporal evolution of China's oases between 2000 and 2020 was analyzed. This was based on the normalized difference vegetation index derived from moderate imaging spectroradiometer data, combined with superpixel image segmentation, the region adjacency graph merging method, and the genetic algorithm. The results were as follows. Between 2000 and 2020: (1) the oases in China expanded significantly from 207,812 to 289,751 km2; with a total increase of 39.43%; and the number of oases decreased from 135 to 93, which mainly occurred in micro and small oases; (2) the areas of oases in stable, expanding, and shrinking states accounted for 96.77%, 42.65%, and respectively 3.23% of the total oasis area in 2000. Oases have become more concentrated and continuously expanded; (3) the increase in glacial meltwater and precipitation caused by rising temperatures increased the availability of water resources in oases, which provides the substantial basis for expansion; and (4) the primary anthropogenic drivers of oasis expansion and stabilization regions were cultivation and construction, respectively.

GeoDLS: A Deep Learning-Based Corn Disease Tracking and Location System Using RTK Geolocated UAS Imagery

Deep learning-based solutions for precision agriculture have recently achieved promising results. Deep learning has been used to identify crop diseases at the initial stages of disease development in an effort to create effective disease management systems. However, the use of deep learning and unmanned aerial system (UAS) imagery to track the spread of diseases, identify diseased regions within cornfields, and notify users with actionable information remains a research gap. Therefore, in this study, high-resolution, UAS-acquired, real-time kinematic (RTK) geotagged, RGB imagery at an altitude of 12 m above ground level (AGL) was used to develop the Geo Disease Location System (GeoDLS), a deep learning-based system for tracking diseased regions in corn fields. UAS images (resolution 8192 × 5460 pixels) were acquired in cornfields located at Purdue University’s Agronomy Center for Research and Education (ACRE), using a DJI Matrice 300 RTK UAS mounted with a 45-megapixel DJI Zenmuse P1 camera during corn stages V14 to R4. A dataset of 5076 images was created by splitting the UAS-acquired images using tile and simple linear iterative clustering (SLIC) segmentation. For tile segmentation, the images were split into tiles of sizes 250 × 250 pixels, 500 × 500 pixels, and 1000 × 1000 pixels, resulting in 1804, 1112, and 570 image tiles, respectively. For SLIC segmentation, 865 and 725 superpixel images were obtained using compactness (m) values of 5 and 10, respectively. Five deep neural network architectures, VGG16, ResNet50, InceptionV3, DenseNet169, and Xception, were trained to identify diseased, healthy, and background regions in corn fields. DenseNet169 identified diseased, healthy, and background regions with the highest testing accuracy of 100.00% when trained on images of tile size 1000 × 1000 pixels. Using a sliding window approach, the trained DenseNet169 model was then used to calculate the percentage of diseased regions present within each UAS image. Finally, the RTK geolocation information for each image was used to update users with the location of diseased regions with an accuracy of within 2 cm through a web application, a smartphone application, and email notifications. The GeoDLS could be a potential tool for an automated disease management system to track the spread of crop diseases, identify diseased regions, and provide actionable information to the users.

Efficient Color Quantization Using Superpixels.

We propose three methods for the color quantization of superpixel images. Prior to the application of each method, the target image is first segmented into a finite number of superpixels by grouping the pixels that are similar in color. The color of a superpixel is given by the arithmetic mean of the colors of all constituent pixels. Following this, the superpixels are quantized using common splitting or clustering methods, such as median cut, k-means, and fuzzy c-means. In this manner, a color palette is generated while the original pixel image undergoes color mapping. The effectiveness of each proposed superpixel method is validated via experimentation using different color images. We compare the proposed methods with state-of-the-art color quantization methods. The results show significantly decreased computation time along with high quality of the quantized images. However, a multi-index evaluation process shows that the image quality is slightly worse than that obtained via pixel methods.

Using geostationary satellite ocean color data and superpixel to map the diurnal dynamics of water transparency in the eastern China seas

Polar-orbiting ocean color satellites can monitor daily to interannual variations in water transparency (or Secchi disk depth, SDD) from regional to global oceans. However, diurnal variations in SDD of coastal oceans remain poorly understood. Based on the bio-optical SDD algorithm, we retrieved the SDD products of hourly observations using the Geostationary Ocean Color Imager (GOCI) from 2011 to 2020 in the eastern China seas. The determination coefficient (R2) between the SDD product and the in situ dataset is 0.93, with a root mean squared error (RMSE) of 0.86 m. Based on the pixel-level tempo-spatial analysis, superpixel image segmentation retrieved by a simple linear iterative clustering algorithm (SLIC) was applied to classify the SDD product. The reconstruction SDD superpixel products not only match the spatial distribution well but can also more clearly express the spatial gradient. The percentage of the diurnal change in transparency was high in the nearshore (∼10 %), medium in transitional waters (∼5%), and low in offshore waters (∼3%). Finally, we found a significant negative correlation between SDD and wind speed (R2 = 0.65) and a significantly positive correlation between diurnal change range (DCR) and wind speed (R2 = 0.62). In contrast, sea surface temperature (SST) was positively correlated with SDD (R2 = 0.72) but significantly negatively correlated with DCR (R2 = 0.80). These results provide a basis for studying diurnal SDD changes in highly dynamic waters.

Generic parallel data structures and algorithms to GPU superpixel image segmentation

We propose parallel implementation on GPU (graphics processing unit) system for some generic algorithms applied to superpixel image segmentation problem. The aim is to provide standard algorithms based on generic decentralized data structures that could be easily improved and customized on many optimization problems on parallel platforms. Note that superpixel segmentation are clustering algorithms applied to image processing. Two types of algorithms are presented and implemented on GPU based on common parallel data structures. Firstly, we present a parallel implementation of the well-known k-means algorithm with application to 3D data. It is based on a cellular grid subdivision of space that allows closest point findings in constant optimal time for bounded distributions. Secondly, we present an application of the parallel Boruvka minimum spanning forest algorithm to compute watershed segmentation. Both techniques are fully executed on GPU and share the same data structures that embed disjoint-set-trees and distributed-link-lists. We evaluate our k-means approach with regards to state-of-the-art methods, that are, the well known SLIC algorithm, and the adaptive segmentation approach SPASM. We argue that our implementation has the shortest execution time among the tested methods, with a near real time performance and quasi linear acceleration factor, while it provides more regular shape superpixel segmentation based on hexagonal tessellation. Watershed minimum spanning forest method is presented and evaluated accordingly to the same experimental framework.

Electrochemical Intelligent Recognition of Mineral Materials Based on Superpixel Image Segmentation.

In order to study the needs of identifying rock thin-section samples by manual observation in the field of geology, a method of electrochemical intelligent recognition of mineral materials based on superpixel image segmentation is proposed. The image histogram of this method can be used to represent the distribution of each pixel value of the image. This interval is consistent with the number of pixels in the method. And using the experiment, the CPU used in the experiment is Intel® Core™ i7-8700 3.2 GHz, the memory is 16 GB, and the GPU is NVIDIA GeForce GT × 1080 Ti, which ensures the accuracy of the experiment. Based on all the experimental results, it can be seen that after the two-stage processing of the designed superpixel algorithm and the region merging algorithm, the final sandstone slice image segmentation results are close to the results of manual labeling, which is helpful for the subsequent research on sandstone component identification. The feasibility of this method was verified.

Pixel Resolution Imaging in Parallel Phase-Shifting Digital Holography

Parallel phase-shifting digital holography (PPSDH) employing a polarization image sensor can suppress zero-order and twin-image noise through a single exposure, achieve instantaneous measurement of complex-valued dynamic objects, and have broad applications in the areas of biomedicine, etc. To improve the imaging resolution of PPSDH, we propose an oversampled super-pixel image reconstruction method, which can be expressed as the implementation of nearest-neighbor interpolation to replace blank pixels in sparse sub-phase-shift holograms. We found experimentally that the maximum spatial lateral resolution of the reconstructed image based on the existing super-pixel method, B-spline, bicubic, bilinear, and the proposed nearest-neighbor interpolation was 12.4 µm, 11.4 µm, 9.8 µm, 8.8 µm, and 7.8 µm, respectively. The main reason for not reaching the ideal value of 6.9 µm was the inherent residual zero-order and twin-image noise, which needs to be removed in the future.

Biomimetic Vision for Zoom Object Detection Based on Improved Vertical Grid Number YOLO Algorithm.

With the development of bionic computer vision for images processing, researchers have easily obtained high-resolution zoom sensing images. The development of drones equipped with high-definition cameras has greatly increased the sample size and image segmentation and target detection are important links during the process of image information. As biomimetic remote sensing images are usually prone to blur distortion and distortion in the imaging, transmission and processing stages, this paper improves the vertical grid number of the YOLO algorithm. Firstly, the light and shade of a high-resolution zoom sensing image were abstracted, and the grey-level cooccurrence matrix extracted feature parameters to quantitatively describe the texture characteristics of the zoom sensing image. The Simple Linear Iterative Clustering (SLIC) superpixel segmentation method was used to achieve the segmentation of light/dark scenes, and the saliency area was obtained. Secondly, a high-resolution zoom sensing image model for segmenting light and dark scenes was established to made the dataset meet the recognition standard. Due to the refraction of the light passing through the lens and other factors, the difference of the contour boundary light and dark value between the target pixel and the background pixel would make it difficult to detect the target, and the pixels of the main part of the separated image would be sharper for edge detection. Thirdly, a YOLO algorithm with an improved vertical grid number was proposed to detect the target in real time on the processed superpixel image array. The adjusted aspect ratio of the target in the remote sensing image modified the number of vertical grids in the YOLO network structure by using 20 convolutional layers and five maximum aggregation layers, which was more accurately adapted to “short and coarse” of the identified object in the information density. Finally, through comparison with the improved algorithm and other mainstream algorithms in different environments, the test results on the aid dataset showed that in the target detection of high spatial resolution zoom sensing images, the algorithm in this paper showed higher accuracy than the YOLO algorithm and had real-time performance and detection accuracy.

Gradient- enhanced waterpixels clustering for coal gangue image segmentation

ABSTRACT The estimation of gangue content based on image analysis is an essential part of the intelligent production of top coal caving. Image segmentation is a very important step before the image analysis. As a pre-processing tool of image segmentation, high-efficient superpixel algorithms have been widely used in many real-time vision applications. In order to effectively use the over segmentation problem of noise image caused by traditional watershed transform, a gradient-enhanced waterpixels fast clustering segmentation algorithm is proposed to obtain more accurate and robust segmentation results of coal gangue images in less computing time. Firstly, the edge gradient features of the gangue image are highlighted through multi-scale detail enhancement. Secondly, the superpixel with accurate contour is formed based on the watershed transform of the gradient image reconstructed by multi-scale morphology (MMR). Finally, based on the obtained superpixel image, the final segmentation result is obtained by calculating the pixel statistical histogram of each region in the super-pixel image and clustering the super-pixel image by using fuzzy c-means (FCM) clustering algorithm. Experiments performed on coal gangue images demonstrate that the proposed algorithm obtains accurate and continuous target contour and reaches the requirement of human visual characteristics. According to the evaluation index of superpixel algorithm, for complex application scenes with high real-time requirements, such as the process of top coal caving, the proposed super-pixel clustering segmentation method has better segmentation effect compared with the most advanced image segmentation algorithm.

Super-Pixel Guided Low-Light Images Enhancement with Features Restoration.

Dealing with low-light images is a challenging problem in the image processing field. A mature low-light enhancement technology will not only be conductive to human visual perception but also lay a solid foundation for the subsequent high-level tasks, such as target detection and image classification. In order to balance the visual effect of the image and the contribution of the subsequent task, this paper proposes utilizing shallow Convolutional Neural Networks (CNNs) as the priori image processing to restore the necessary image feature information, which is followed by super-pixel image segmentation to obtain image regions with similar colors and brightness and, finally, the Attentive Neural Processes (ANPs) network to find its local enhancement function on each super-pixel to further restore features and details. Through extensive experiments on the synthesized low-light image and the real low-light image, the experimental results of our algorithm reach 23.402, 0.920, and 2.2490 for Peak Signal to Noise Ratio (PSNR), Structural Similarity (SSIM), and Natural Image Quality Evaluator (NIQE), respectively. As demonstrated by the experiments on image Scale-Invariant Feature Transform (SIFT) feature detection and subsequent target detection, the results of our approach achieve excellent results in visual effect and image features.

A hybrid method for traumatic brain injury lesion segmentation

<span>Traumatic brain injuries are significant effects of disability and loss of life. Physicians employ computed tomography (CT) images to observe the trauma and measure its severity for diagnosis and treatment. Due to the overlap of hemorrhage and normal brain tissues, segmentation methods sometimes lead to false results. The study is more challenging to unitize the AI field to collect brain hemorrhage by involving patient datasets employing CT scans images. We propose a novel technique free-form object model for brain injury CT image segmentation based on superpixel image processing that uses CT to analyzing brain injuries, quite challenging to create a high outstanding simple linear iterative clustering (SLIC) method. The maintains a strategic distance of the segmentation image to reduced intensity boundaries. The segmentation image contains marked red hemorrhage to modify the free-form object model. The contour labelled by the red mark is the output from our free-form object model. We proposed a hybrid image segmentation approach based on the combined edge detection and dilation technique features. The approach diminishes computational costs, and the show accomplished 96.68% accuracy. The segmenting brain hemorrhage images are achieved in the clustered region to construct a free-form object model. The study also presents further directions on future research in this domain.</span>

Study on TCM Tongue Image Segmentation Model Based on Convolutional Neural Network Fused with Superpixel.

Tongue image segmentation is a base work of TCM tongue processing. Nowadays, deep learning methods are widely used on tongue segmentation, which has better performance than conventional methods. However, when the tongue color is close to the color of the adjoining area, the contour of tongue segmentation by deep learning may be coarse which could influence the subsequent analysis. Here a novel tongue image segmentation model based on a convolutional neural network fused with superpixel was proposed to solve the problem. Methods. On the basis of a convolutional neural network fused with superpixel, the novel tongue image segmentation model SpurNet was proposed in this study. The residual structure of ResNet18 was introduced as the feature extraction layer on the encoding path, to construct the first stage processing module UrNet of SpurNet. The superpixel segmentation was fused with UrNet to form the second stage process of SpurNet. To verify the effect of SpurNet. The models before and after fusion with superpixel, classical image segmentation models FCN and DeepLab were compared with SpurNet on the dataset of 367 manually labeled tongue images. Results. The SpurNet model performance test with 10-fold cross-validation showed PA of 0.9145 ± 0.0043, MPA of 0.9168 ± 0.0048, MIoU of 0.8417 ± 0.0072 and FWIoU of 0.8454 ± 0.0072. Relative to FCN, DeepLab and their superpixel fused models, the SpurNet model was superior in tongue image segmentation and could increase PA by 1.91%–3.17%, MPA by 1.38%–2.61%, MIoU by 3.09%–5.07%, and FWIoU by 3.11%–5.08%. Compared to UrNet, the first stage processing module, the SpurNet model also increased the PA, MPA, MIoU and FWIoU by 0.15%, 0.09%, 0.24% and 0.24%, respectively. Conclusion. The SpurNet model, after fusing with superpixel image segmentation, can better accomplish the task of tongue image segmentation, more accurately process the margins of tongue and resolve the over-segmentation and under-segmentation. The thought of this study is a new exploration in the field of tongue image segmentation, which could provide a reference for the modern research on TCM tongue images.

Proxy-Bridged Image Reconstruction Network for Anomaly Detection in Medical Images.

Anomaly detection in medical images refers to the identification of abnormal images with only normal images in the training set. Most existing methods solve this problem with a self-reconstruction framework, which tends to learn an identity mapping and reduces the sensitivity to anomalies. To mitigate this problem, in this paper, we propose a novel Proxy-bridged Image Reconstruction Network (ProxyAno) for anomaly detection in medical images. Specifically, we use an intermediate proxy to bridge the input image and the reconstructed image. We study different proxy types, and we find that the superpixel-image (SI) is the best one. We set all pixels' intensities within each superpixel as their average intensity, and denote this image as SI. The proposed ProxyAno consists of two modules, a Proxy Extraction Module and an Image Reconstruction Module. In the Proxy Extraction Module, a memory is introduced to memorize the feature correspondence for normal image to its corresponding SI, while the memorized correspondence does not apply to the abnormal images, which leads to the information loss for abnormal image and facilitates the anomaly detection. In the Image Reconstruction Module, we map an SI to its reconstructed image. Further, we crop a patch from the image and paste it on the normal SI to mimic the anomalies, and enforce the network to reconstruct the normal image even with the pseudo abnormal SI. In this way, our network enlarges the reconstruction error for anomalies. Extensive experiments on brain MR images, retinal OCT images and retinal fundus images verify the effectiveness of our method for both image-level and pixel-level anomaly detection.

Segmentation of Rat Brains and Cerebral Hemispheres in Triphenyltetrazolium Chloride-Stained Images after Stroke.

Ischemic stroke is one of the leading causes of death among the aged population in the world. Experimental stroke models with rodents play a fundamental role in the investigation of the mechanism and impairment of cerebral ischemia. For its celerity and veracity, the 2,3,5-triphenyltetrazolium chloride (TTC) staining of rat brains has been extensively adopted to visualize the infarction, which is subsequently photographed for further processing. Two important tasks are to segment the brain regions and to compute the midline that separates the brain. This paper investigates automatic brain extraction and hemisphere segmentation algorithms in camera-based TTC-stained rat images. For rat brain extraction, a saliency region detection scheme on a superpixel image is exploited to extract the brain regions from the raw complicated image. Subsequently, the initial brain slices are refined using a parametric deformable model associated with color image transformation. For rat hemisphere segmentation, open curve evolution guided by the gradient vector flow in a medial subimage is developed to compute the midline. A wide variety of TTC-stained rat brain images captured by a smartphone were produced and utilized to evaluate the proposed segmentation frameworks. Experimental results on the segmentation of rat brains and cerebral hemispheres indicated that the developed schemes achieved high accuracy with average Dice scores of 92.33% and 97.15%, respectively. The established segmentation algorithms are believed to be potential and beneficial to facilitate experimental stroke study with TTC-stained rat brain images.

Superpixel image segmentation of VISTA expression in colorectal cancer and its relationship to the tumoral microenvironment

Colorectal cancer (CRC) is the third most common cause of cancer related death in the United States (Jasperson et al. in Gastroenterology 138:2044–2058, https://doi.org/10.1053/j.gastro.2010.01.054, 2010). Many studies have explored prognostic factors in CRC. Today, much focus has been placed on the tumor microenvironment, including different immune cells and the extracellular matrix (ECM). The present study aims to evaluate the role of V-domain immunoglobulin suppressor of T cell activation (VISTA). We utilized QuPath for whole slides image analysis, performing superpixel image segmentation (SIS) on a 226 patient-cohort. High VISTA expression correlated with better disease-free survival (DFS), high tumor infiltrative lymphocyte, microsatellite instability, BRAF mutational status as well as lower tumor stage. High VISTA expression was also associated with mature stromal differentiation (SD). When cohorts were separated based on SD and MMR, only patients with immature SD and microsatellite stability were found to correlate VISTA expression with DFS. Considering raised VISTA expression is associated with improved survival, TILs, mature SD, and MMR in CRC; careful, well-designed clinical trials should be pursued which incorporate the underlying tumoral microenvironment.

MURAL LINEAR FEATURE ENHANCEMENT ALGORITHM COMBINING SEMI-SUPERVISED IMAGE SEGMENTATION AND DIMENSIONALITY REDUCTION

Abstract. Due to the influence of natural and human factors, the linear features in the murals are partially blurred, which brings great challenges to the digital preservation and virtual restoration of cultural heritage. Taking the advantages of non-invasive measurement as well as the rich image and spectral information of hyperspectral technology, we proposed a linear feature enhancement method by combining semi-supervised superpixel segmentation with block dimension reduction. The main research work includes: (1) The true color composite image was segmented to obtain the label data by using the local spatial information of the superpixel image and the global feature information extracted by fuzzy c-means (FCM) clustering.(2) According to the label data, the preprocessed hyperspectral data were divided into homogeneous regions, whose dimensionality was reduced by principal component analysis (PCA) and kernel principal component analysis (KPCA). (3) The principal component images with the largest gradient after dimensionality reduction were respectively selected and normalized. The optimal principal component images normalized by the block PCA and block KPCA dimensionality reduction algorithms are superimposed to produce the linear feature enhancement images of murals. The hyperspectral images of some murals in Qutan Temple, Qinghai Province, China were used to verify the method. The results show that the spatial information and the spectral information of different pattern areas in the hyperspectral image can be fully used by combining the superpixel FCM image segmentation algorithm with the dimensionality reduction algorithm. of. It can highlight the linear information in the hyperspectral images of fades murals.

Breast ultrasound image segmentation: A coarse-to-fine fusion convolutional neural network.

Breast ultrasound (BUS) image segmentation plays a crucial role in computer-aided diagnosis systems for BUS examination, which are useful for improved accuracy of breast cancer diagnosis. However, such performance remains a challenging task owing to the poor image quality and large variations in the sizes, shapes, and locations of breast lesions. In this paper, we propose a new convolutional neural network with coarse-to-fine feature fusion to address the aforementioned challenges. The proposed fusion network consists of an encoder path, a decoder path, and a core fusion stream path (FSP). The encoder path is used to capture the context information, and the decoder path is used for localization prediction. The FSP is designed to generate beneficial aggregate feature representations (i.e., various-sized lesion features, aggregated coarse-to-fine information, and high-resolution edge characteristics) from the encoder and decoder paths, which are eventually used for accurate breast lesion segmentation. To better retain the boundary information and alleviate the effect of image noise, we input the superpixel image along with the original image to the fusion network. Furthermore, a weighted-balanced loss function was designed to address the problem of lesion regions having different sizes. We then conducted exhaustive experiments on three public BUS datasets to evaluate the proposed network. The proposed method outperformed state-of-the-art (SOTA) segmentation methods on the three public BUS datasets, with average dice similarity coefficients of 84.71(±1.07), 83.76(±0.83), and 86.52(±1.52), average intersection-over-union values of 76.34(±1.50), 75.70(±0.98), and 77.86(±2.07), average sensitivities of 86.66(±1.82), 85.21(±1.98), and 87.21(±2.51), average specificities of 97.92(±0.46), 98.57(±0.19), and 99.42(±0.21), and average accuracies of 95.89(±0.57), 97.17(±0.3), and 98.51(±0.3). The proposed fusion network could effectively segment lesions from BUS images, thereby presenting a new feature fusion strategy to handle challenging task of segmentation, while outperforming the SOTA segmentation methods. The code is publicly available at https://github.com/mniwk/CF2-NET.