Superpixel Segmentation Algorithm Research Articles

Deep Auto Encoder Based Extreme Learning System for Automatic Segmentation of Cervical Cells

Nowadays, cervical cancer has emerged as one of the major causes of death and incredibly difficult to identify. In this research, three novel methods have been developed for the automatic detection and classification of cervical cancer in Pap images. There are five steps in this framework which includes data collection, preprocessing, segmentation, feature extraction, and classification of images. The proposed method is implemented using Herlev Pap smear data set. The input image from the data set is pre-processed by using an Anisotropic Diffusion Filter with Unsharp Masking technique that removes the noise in the Pap data. In the next step, an enhanced image sequence is segmented automatically by the proposed Advance Map-Based Superpixel Segmentation (AMBSS) algorithm. Finally, cervix cancer images are procured by utilizing an AMBSS and classified by Support Vector Machine classifier. The accuracy obtained is 85.4%. In order to improve the accuracy AMBSS with quasi newton-based Feed Forward Neural Network classification is used and the accuracy of 96.0% is obtained. Furthermore, AMBSS with Deep auto encoder-based Extreme Learning Machine classification is performed and achieved the accuracy of 99.1%. The findings show that the precision of classification typically exceeds other intelligent methods previously applied.

A New Clustering Algorithm Toward Building Segmentation From Aerial Images by Utilizing RGB‐Component Differences

AbstractFor the image segmentation task, deep features‐based deep learning methods have strict data requirements on the quality and quantity, which limits their application in irregular scenes, such as the villages subjected to geological hazards in the complex mountains. However, traditional algorithms with shallow and middle features are unable to utilize the complex information of unmanned aerial vehicle (UAV) remote sensing images effectively. To address this issue, we proposed a new image segmentation algorithm by using the RGB component difference clustering (RGBCDC). In this method, after dimensionality reduction of RGB color space from three dimensions to two dimensions, the distance between similar objects would be diminished and the distance between disturbance areas would be enlarged. Two different UAV data sets (3,000 in total) have been used to compare the proposed algorithm with the traditional segmentation algorithms and the deep learning‐based methods. Results show that the average pixel accuracy is 13.11%, 11.86%, 13.84%, 8.00%, and 9.86% higher than Octree Quantization clustering algorithm, Region Growing algorithm, Watershed algorithm, K‐Means algorithm, and SLIC superpixel segmentation algorithm, respectively. When facing insufficient samples, the new method performed better than deep learning algorithms, such as Deeplabv3, PSPnet, UNET, and UISBB. In general, the proposed algorithm for building segmentation shows a better potential for use in uncommon situations, especially for rapid emergency rescue after serious mountain hazards.

Cluster-based fine-to-coarse superpixel segmentation

As an image preprocessing technology, superpixel segmentation has become an important tool in the field of computer vision. How to obtain a more accurate, faster, and easier-to-apply superpixel segmentation algorithm is a problem faced by researchers. In this paper, a cluster-based fine-to-coarse superpixel segmentation (FCSS) algorithm is proposed. By introducing color thresholds and depth thresholds with practical physical meanings as algorithm parameters, high-quality segmentation with fewer superpixels is achieved. It not only reduces the complexity of the upper application, but also provides an easy to understand interface. Superpixel segmentation methods often cannot achieve high-quality segmentation through a set of parameters. Experimental results show that FCSS can achieve finer segmentation by setting different parameters, and the segmentation results are superior to other algorithms. When the number of superpixels is 100, the segmentation performance of FCSS is better than that of existing state-of-the-art methods.

A Method for Human Burn Diagnosis Using Machine Learning and SLIC Superpixels Based Segmentation

The burn is a global health issue. The treatment of burn is performed by a burn surgeon. Burn region color is used to categorize burn types by a burn surgeon. Due to the lack of resources in many places burn patient dies. In such a situation a computer-based diagnosis system is very helpful. The severity of the burn is based on the depth and size of the burn region. The burn severity analysis can be accurately done by using the CIELab image as it contains all spectrums of colors. In the proposed work, the CIELab image is segmented using SLIC (Simple Linear Iterative Clustering) superpixel segmentation algorithm to form the same color pixels clusters. These clusters are used to diagnosis the severity of a burn, The severity of a burn is measured by the value of kurtosis. A high kurtosis value implies a larger depth and a small value implies a smaller depth. Based on the depth of burn, grafting or non-grafting is suggested by a burn surgeon. We observed depth of the burn is high in full-thickness burn compared to the deep dermal and the superficial dermal burn. The proposed method can correctly identify full-thickness burn and most of the deep dermal burn image. The open-access standard data set BURNS BIP-US DATABASE is used for training and testing of the proposed model on SVM (Support Vector Machine). The proposed model can classify with an accuracy of 89.16% and F1- score 90%. The achieved accuracy of the model is better than 79% of multidimensional scaling (MDS) in previous work.

CONIC: Contour Optimized Non-Iterative Clustering Superpixel Segmentation

Superpixels group perceptually similar pixels into homogeneous sub-regions that act as meaningful features for advanced tasks. However, there is still a contradiction between color homogeneity and shape regularity in existing algorithms, which hinders their performance in further processing. In this work, a novel Contour Optimized Non-Iterative Clustering (CONIC) method is presented. It incorporates contour prior into the non-iterative clustering framework, aiming to provide a balanced trade-off between segmentation accuracy and visual uniformity. After the conventional grid sampling initialization, a regional inter-seed correlation is first established by the joint color-spatial-contour distance. It then guides a global redistribution of all seeds to modify the number and positions iteratively. This is done to avoid clustering falling into the local optimum and achieve the exact number of user-expectation. During the clustering process, an improved feature distance is elaborated to measure the color similarity that considers contour constraint and prevents the boundary pixels from being wrongly assigned. Consequently, superpixels acquire better visual quality and their boundaries are more consistent with the object contours. Experimental results show that CONIC performs as well as or even better than the state-of-the-art superpixel segmentation algorithms, in terms of both efficiency and segmentation effects.

Multiscale Weighted Adjacent Superpixel-Based Composite Kernel for Hyperspectral Image Classification

This paper presents a composite kernel method (MWASCK) based on multiscale weighted adjacent superpixels (ASs) to classify hyperspectral image (HSI). The MWASCK adequately exploits spatial-spectral features of weighted adjacent superpixels to guarantee that more accurate spectral features can be extracted. Firstly, we use a superpixel segmentation algorithm to divide HSI into multiple superpixels. Secondly, the similarities between each target superpixel and its ASs are calculated to construct the spatial features. Finally, a weighted AS-based composite kernel (WASCK) method for HSI classification is proposed. In order to avoid seeking for the optimal superpixel scale and fuse the multiscale spatial features, the MWASCK method uses multiscale weighted superpixel neighbor information. Experiments from two real HSIs indicate that superior performance of the WASCK and MWASCK methods compared with some popular classification methods.

GPU accelerated waterpixel algorithm for superpixel segmentation of hyperspectral images

The high computational cost of the superpixel segmentation algorithms for hyperspectral remote sensing images makes them ideal candidates for parallel computation. The waterpixel algorithm, in particular, extracts segmentation regions called waterpixels and consists of four stages called vectorial gradient, spatial regularization, marker selection, and watershed transform. In this paper, an efficient version of a GPU algorithm for waterpixel segmentation using the Compute Unified Device Architecture (CUDA) is presented. The algorithm extracts all the spectral information available in the bands of the hyperspectral image through the vectorial gradient. A cellular automaton is selected for the computation of the watershed transform using a block-asynchronous implementation with 8-connectivity. The experimental analysis shows high speedup values for the resulting GPU algorithm when it is compared to a multicore OpenMP implementation using 8 threads.

Flexible Gabor-Based Superpixel-Level Unsupervised LDA for Hyperspectral Image Classification

Hyperspectral images encompass abundant information and provide unique characteristics for material classification. However, the labeling of training samples can be challenging in hyperspectral image classification. To address this problem, this study proposes a framework named flexible Gabor-based superpixel-level unsupervised linear discriminant analysis (FG- <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Su</i> ULDA) to extract the most informative and discriminating features for classification. First, a number of 3-D flexible Gabor filters are rigorously designed using an asymmetric sinusoidal wave to sufficiently characterize the spatial–spectral structure in hyperspectral images. Then, an unsupervised linear discriminant analysis strategy guided by the entropy rate superpixel (ERS) segmentation algorithm, called <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Su</i> ULDA, is skillfully introduced to reduce the extracted large amount of FG features. The <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Su</i> ULDA method not only boosts the classification capability but also increases the peculiarity of features, with the aid of superpixel information. Finally, the achieved features are imported to the popular support vector machine classifier. The proposed FG- <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Su</i> ULDA framework is applied to four real hyperspectral image data sets, and the experiments constantly prove that our FG- <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Su</i> ULDA is superior to several state-of-the-art methods in both classification performance and computational efficiency, especially with scarce training samples. The codes of this work are available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">http://jiasen.tech/papers/</uri> for the sake of reproducibility.

Spectral-Spatial Active Learning With Structure Density for Hyperspectral Classification

In this paper, a spectral-spatial active learning (AL) method is proposed based on an up-to-date unlabeled samples sampling strategy concentrated on the structure density supported by breaking ties. The proposed sampling criterion in AL is used for hyperspectral image classification, which involves several steps: First, superpixel segmentation algorithm is conducted on the HSI to cluster pixels with similar spectral-spatial signature into the same superpixel block. Then, density peak clustering technique is performed on the each superpixel block to obtain structure density of the pixels. Meanwhile, probability-based classifier is employed to achieve the probability distributions of pixel. Next, breaking ties (BT) score of each pixel can be calculated by exploiting the probabilities. Additionally, a fusion mechanism is introduced to select the unlabeled samples with representativeness and informativeness advantages by employing the BT-assisted structure density (SD sampling criterion) of each pixel. Finally, the samples with manual labeled class labels are put into the training set to retrain the classifier. Experimental results manifest that the proposed SD-based sampling criterion in active learning can significantly improve the classification accuracy in few labor costs. Thus, it has certain feasibility in practical application.

Adaptive Superpixel Segmentation of Marine SAR Images by Aggregating Fisher Vectors

Superpixel segmentation is an important technique for image analysis. In this article, we develop a new superpixel segmentation approach and investigate its application on ship target detection in marine synthetic aperture radar (SAR) images. Existing superpixel segmentation algorithms often simply consider the intensity and spatial features, which may degrade the segmentation performance due to the low contrast between ship targets and the sea clutter background in marine SAR images. Besides, it is difficult for existing algorithms to adaptively select the weights of the features. Here, we propose a new Fisher vector (FV)-based adaptive superpixel segmentation (FVASS) algorithm to address the aforementioned issues. Our newly developed FVASS not only fuses the intensity and spatial features, but also the multiorder features introduced by FVs, resulting in a better segmentation performance (even with low signal-to-clutter ratios). The weights of the features considered in FVASS are adaptively adjusted by minimizing the sum of within-superpixel variances to maintain the compactness of superpixels. Experiments demonstrate that, compared with commonly used superpixel segmentation methods, the proposed FVASS algorithm enhances the segmentation performance of SAR images and further improves the detection performance of existing superpixel-based ship detectors.

Novel method for liver cancer detection using super pixel and ANN algorithm

Nowadays most of the people are affected by liver cancer that can be the leading cause of cancer death. To analyze the liver cancer detection using super pixel segmentation and artificial neural network (ANN) Algorithm and it reduces the time complexity. An automatic 3D segmentation method for high energy sound waves in ultrasound liver images, which is based on a fusion of accurate time-consuming method and statistical texture prior. Three orthogonal linear filter banks are used to extract texture features from each image in the region around the point of the database. The atlas database includes registered liver images from previous patients and their segmented liver surfaces. To train the problem of patient–specific frequency content in the image features from the atlas database are using ANN algorithm and then to segment the super pixel segmentation the liver image from a new patient. The dataset of liver cancer detection are benign and malign.

Shearlet-Based Structure-Aware Filtering for Hyperspectral and LiDAR Data Classification

The joint interpretation of hyperspectral images (HSIs) and light detection and ranging (LiDAR) data has developed rapidly in recent years due to continuously evolving image processing technology. Nowadays, most feature extraction methods are carried out by convolving the raw data with fixed-size filters, whereas the structural and texture information of objects in multiple scales cannot be sufficiently exploited. In this article, a shearlet-based structure-aware filtering approach, abbreviated as ShearSAF, is proposed for HSI and LiDAR feature extraction and classification. Specifically, superpixel-guided kernel principal component analysis (KPCA) is firstly adopted on raw HSIs to reduce the dimensions. Then, the KPCA-reduced HSI and LiDAR data are converted to the shearlet domain for texture and area feature extraction. In contrast, superpixel segmentation algorithm utilizes the raw HSI data to obtain the initial oversegmentation map. Subsequently, by utilizing a well-designed minimum merging cost that fully considers spectral (HSI and LiDAR data), texture, and area features, a region merging procedure is gradually conducted to produce a final merging map. Further, a scale map that locally indicates the filter size is achieved by calculating the edge distance. Finally, the KPCA-reduced HSI and LiDAR data are convolved with the locally adaptive filters for feature extraction, and a random forest (RF) classifier is thus adopted for classification. The effectiveness of our ShearSAF approach is verified on three real-world datasets, and the results show that the performance of ShearSAF can achieve an accuracy higher than that of comparison methods when exploiting small-size training sample problems. The codes of this work will be available at http://jiasen.tech/papers/ for the sake of reproducibility.

Convex and Compact Superpixels by Edge- Constrained Centroidal Power Diagram.

Superpixel segmentation, as a central image processing task, has many applications in computer vision and computer graphics. Boundary alignment and shape compactness are leading indicators to evaluate a superpixel segmentation algorithm. Furthermore, convexity can make superpixels reflect more geometric structures in images and provide a more concise over-segmentation result. In this paper, we consider generating convex and compact superpixels while satisfying the constraints of adhering to the boundary as far as possible. We formulate the new superpixel segmentation into an edge-constrained centroidal power diagram (ECCPD) optimization problem. In the implementation, we optimize the superpixel configurations by repeatedly performing two alternative operations, which include site location updating and weight updating through a weight function defined by image features. Compared with existing superpixel methods, our method can partition an image into fully convex and compact superpixels with better boundary adherence. Extensive experimental results show that our approach outperforms existing superpixel segmentation methods in boundary alignment and compactness for generating convex superpixels.

A new deep learning approach for the retinal hard exudates detection based on superpixel multi-feature extraction and patch-based CNN

Diabetic Retinopathy (DR) is a severe complication of chronic diabetes causing significant visual deterioration and may lead to blindness with delay of being treated. Exudative diabetic maculopathy, a form of macular edema where hard exudates (HE) develop, is a frequent cause of visual deterioration in DR. The detection of HE comprises a significant role in the DR diagnosis. In this paper, an automatic exudates detection method based on superpixel multi-feature extraction and patch-based deep convolutional neural network is proposed. Firstly, superpixels, regarded as candidates, are generated on each resized image using the superpixel segmentation algorithm called Simple Linear Iterative Clustering (SLIC). Then, 25 features extracted from resized images and patches are generated on each feature. Patches are subsequently used to train a deep convolutional neural network, which distinguishes the hard exudates from the background. Experiments conducted on three publicly available datasets (DiaretDB1, e-ophtha EX and IDRiD) demonstrate that our proposed methodology achieved superior HE detection when compared with current state-of-art algorithms.

Multilevel Information Fusion-Based Change Detection for Multiangle PolSAR Images

Change detection is a key technology in the field of polarimetric synthetic aperture radar (PolSAR) image processing. The current research on the change detection mainly focuses on studying PolSAR images with the same angle or small angle difference, and the angle problem is not considered. However, when the angle difference occurs, especially a large angle difference, some pixels might be falsely detected because the angle difference can affect the polarimetric characteristics. In this letter, we propose a multilevel information fusion-based (MIFB) method, which is suitable for extracting change information from PolSAR images with angle difference. In particular, the proposed method first adopts data resolution correction, then applies an improved feature-based registration algorithm, and finally, incorporates weighted graph theory with the superpixel segmentation algorithm to extract and merge pixel-based and object-based change areas to eliminate false alarms. Experimental results for multitemporal and multiangle PolSAR images reveal that the MIFB method can effectively eliminate false detection caused by angle differences and improve the detection accuracy.

Texture-aware and structure-preserving superpixel segmentation

Superpixel segmentation is an important pre-processing step in many computer vision applications. However, existing methods are mainly constructed on the basis of pixel-level features, which are generally vulnerable to the interference of strong gradient textures and fail to make the generated superpixels adhere to the object boundary well. Therefore, we bring forward a texture-aware and structure-preserving superpixel segmentation algorithm. First, we propose a cluster centroid oriented quarter-circular mask, which prevents the fine-scale structure pixels from sampling inconsistent texture pixels when gathering neighboring information from the local image patch. Next, we put forward a texture-suppressed and structure-preserving hybrid gradient. Its magnitude can well represent the probability that a pixel belongs to the structure one, and its direction can embody the main direction of image structures. Then, we design a color distance with texture-structure tradeoff strategy, which is based on the quarter-circular mask and the hybrid gradient direction. Finally, we devise an integrated structure-avoiding clustering distance, which can be obtained by seeking the maximum hybrid gradient magnitude along the linear path between a pixel and the superpixel centroid. It can effectively prevent the superpixel from crossing the object boundaries. Experimental results show that the superpixels generated by our algorithm not only adhere to object boundaries closely but also preserve regular shape. Our algorithm outperforms state-of-the-art methods, especially for the images with strong gradient textures.

Image clustering algorithm using superpixel segmentation and non‐symmetric Gaussian–Cauchy mixture model

In this study, an unsupervised clustering algorithm is proposed to label superpixel density images. Firstly, the authors propose a novel superpixel segmentation algorithm driven by a modified fuzzy C-means objective function, Kullback–Leibler (KL) divergence, and an entropy term, which generate superpixels with good boundary adherence and intensity homogeneity. In this model, the logarithm of Gaussian distribution as a new distance metric is used to improve the accuracy of boundary pixel classification, the KL divergence is applied to regularise the fuzzy objective function. Based on this model, the generated superpixel intensity images with a highly distinctive background colour from the colour of the target are obtained. Grouping cues generated by superpixels can affect the performance of image clustering greatly. Next, according to the small amount of clustering data generated by the superpixel intensity images, they construct a non-symmetric mixture model based on a mixture of Gaussian distribution and Cauchy distribution for implementing image clustering. Thus, clustering of colour images is transformed into clustering of these newly generated data. The advantage of this model is its well adaption to different shapes of observed data. Experimental results on publicly available data sets are provided to demonstrate the effectiveness of the proposed algorithm.

An Integrated Object and Machine Learning Approach for Tree Canopy Extraction from UAV Datasets

Unmanned aerial vehicles (UAV) have emerged as new platforms for acquiring ultra-high resolution images, which are challenging for extraction of features using conventional image processing approaches. Tree canopies are required to be constantly monitored for better planning and management. UAV is currently one way to survey canopies over a large area for precisely estimating their geometry. Conventional segmentation techniques are extensively used for image feature extraction. However, they lack in accuracy and require high computational processing when used for ultra-high resolution UAV datasets. These issues can be handled by superpixel segmentation algorithms which have good boundary adherence and are computationally efficient. Simple linear iterative clustering (SLIC) is a subset of superpixel segmentation technique which uses minimum tuning parameters making it most efficient. As the random forest is known for handling multiple attributes and robustness, it can be used for classifying and extracting features from segmented image generated using SLIC. The present study mainly focuses on the automation for extraction of tree canopies along with their object-based attributes from the UAV dataset. The data acquisition was carried out using Trimble UX5 fixed-wing UAV which was further orthorectified at a spatial resolution of 13 cm. The ortho-image was further segmented using SLIC algorithm. Canopy segments are then identified and classified using random forest, which is then merged into trees objects on the basis of their proximity. Accuracy assessment was then carried out for extracted tree canopies and was found that the aforesaid approach could achieve 93% similarity index. The current study highlights the potential of using SLIC segmentation and random forest classification method for tree canopy extraction from the ultra-high resolution ortho-image derived from UAV platforms.

Oil seal surface defect detection using superpixel segmentation and circumferential difference

Surface defects affect the quality and safety of oil seals. It is a challenge to detect such defects in a vision system because of the unequal reflection property of oil seal surfaces and low contrast between the defect and the background. This article proposes a visual detection method (VDM) for oil seal surface defects and outlines two key issues of VDMs. First, we present a superpixel segmentation algorithm based on the significant gray level variation in the radial direction of an oil seal surface image. This image is then divided into several ring belts. Subsequently, considering the reflection inequality and low contrast, we propose a new circumferential background difference algorithm based on the small variation along the circumferential direction of the image. This algorithm eliminates the influence of the reflection inequality and improves the contrast distinction between the defects and the background. The experimental results verify the effectiveness of the proposed method with a recall and precision as high as 95.2% and 86.8%, respectively.

Hyperspectral Imagery Classification Based on Multiscale Superpixel-Level Constraint Representation

Sparse representation (SR)-based models have been widely applied for hyperspectral image classification. In our previously established constraint representation (CR) model, we exploited the underlying significance of the sparse coefficient and proposed the participation degree (PD) to represent the contribution of the training sample in representing the testing pixel. However, the spatial variants of the original residual error-driven frameworks often suffer the obstacles to optimization due to the strong constraints. In this paper, based on the object-based image classification (OBIC) framework, we firstly propose a spectral–spatial classification method, called superpixel-level constraint representation (SPCR). Firstly, it uses the PD in respect to the sparse coefficient from CR model. Then, transforming the individual PD to a united activity degree (UAD)-driven mechanism via a spatial constraint generated by the superpixel segmentation algorithm. The final classification is determined based on the UAD-driven mechanism. Considering that the SPCR is susceptible to the segmentation scale, an improved multiscale superpixel-level constraint representation (MSPCR) is further proposed through the decision fusion process of SPCR at different scales. The SPCR method is firstly performed at each scale, and the final category of the testing pixel is determined by the maximum number of the predicated labels among the classification results at each scale. Experimental results on four real hyperspectral datasets including a GF-5 satellite data verified the efficiency and practicability of the two proposed methods.