Fuzzy segmentation for object‐based image classification

This Letter proposes an object‐based image classification procedure which is based on fuzzy image‐regions instead of crisp image‐objects. The approach has three stages: (a) fuzzification in which fuzzy image‐regions are developed, resulting in a set of images whose digital values express the degree of membership of each pixel to target land‐cover classes; (b) feature analysis in which contextual properties of fuzzy image‐regions are quantified; and (c) defuzzification in which fuzzy image‐regions are allocated to target land‐cover classes. The proposed procedure is implemented using automated statistical techniques that require very little user interaction. The results indicate that fuzzy segmentation‐based methods produce acceptable thematic accuracy and could represent a viable alternative to current crisp image segmentation approaches.     

Satellite‐based precipitation estimation using watershed segmentation and growing hierarchical self‐organizing map

This paper outlines the development of a multi‐satellite precipitation estimation methodology that draws on techniques from machine learning and morphology to produce high‐resolution, short‐duration rainfall estimates in an automated fashion. First, cloud systems are identified from geostationary infrared imagery using morphology based watershed segmentation algorithm. Second, a novel pattern recognition technique, growing hierarchical self‐organizing map (GHSOM), is used to classify clouds into a number of clusters with hierarchical architecture. Finally, each cloud cluster is associated with co‐registered passive microwave rainfall observations through a cumulative histogram matching approach. The network was initially trained using remotely sensed geostationary infrared satellite imagery and hourly ground‐radar data in lieu of a dense constellation of polar‐orbiting spacecraft such as the proposed global precipitation measurement (GPM) mission. Ground‐radar and gauge rainfall measurements were used to evaluate this technique for both warm (June 2004) and cold seasons (December 2004–February 2005) at various temporal (daily and monthly) and spatial (0.04° and 0.25°) scales. Significant improvements of estimation accuracy are found classifying the clouds into hierarchical sub‐layers rather than a single layer. Furthermore, 2‐year (2003–2004) satellite rainfall estimates generated by the current algorithm were compared with gauge‐corrected Stage IV radar rainfall at various time scales over continental United States. This study demonstrates the usefulness of the watershed segmentation and the GHSOM in satellite‐based rainfall estimations.     

Building‐based damage detection due to earthquake using the watershed segmentation of the post‐event aerial images

This paper presents an approach for detecting the damaged buildings due to earthquake using the watershed segmentation of the post‐event aerial images. The approach utilizes the relationship between the buildings and their cast shadows. It is based on an idea that if a building is damaged, it will not produce shadows. The cast shadows of the buildings are detected through an immersion‐based watershed segmentation. The boundaries of the buildings are available and stored in a GIS as vector polygons. The vector‐building boundaries are used to match the shadow casting edges of the buildings with their corresponding shadows and to perform assessments on a building‐specific manner. For each building, a final decision on the damage condition is taken, based on the assessments carried out for that building only. The approach was implemented in Golcuk, one of the urban areas most strongly hit by the 1999 Izmit, Turkey earthquake. To implement the approach, a system called the Building‐Based Earthquake Damage Assessment System was developed in MATLAB. Of the 284 buildings processed and analysed, 229 were correctly labelled as damaged and undamaged, providing an overall accuracy of 80.63%.     

Multiscale stochastic hierarchical image segmentation by spectral clustering

This paper proposes a sampling based hierarchical approach for solving the computational demands of the spectral clustering methods when applied to the problem of image segmentation. The authors first define the distance between a pixel and a cluster, and then derive a new theorem to estimate the number of samples needed for clustering. Finally, by introducing a scale parameter into the simi- larity function, a novel spectral clustering based image segmentation method has been developed. An important characteristic of the approach is that in the course of image segmentation one needs not only to tune the scale parameter to merge the small size clusters or split the large size clusters but also take samples from the data set at the different scales. The multiscale and stochastic nature makes it feasible to apply the method to very large grouping problem. In addition, it also makes the segmentation compute in time that is linear in the size of the image. The experimental results on various synthetic and real world images show the effective- ness of the approach.     

A patch‐based image classification by integrating hyperspectral data with GIS

Hyperspectral remote sensing data provide detailed spectral information and are widely used for pixel‐based image classification. However, without considering spatial correlation among neighbouring pixels, a generated thematic map may have a ‘salt‐and‐pepper’ appearance. With the development of the Geographic Information System (GIS), the spatial relationship between a pixel and its neighbours can be recorded readily and used together with remote sensing data. The objective of this study was to integrate hyperspectral data with the GIS for effective thematic mapping. To date, GIS data have been used mainly in field surveys or training field selection for remote sensing data interpretation. Here we propose a patch‐classification based on integration of the GIS with remote sensing data. The classification results obtained by using this method can be easily saved in a vector format as used for GIS files. Computational cost is decreased compared with a pixel‐by‐pixel classification. The issue of how to identify pure or mixed patches is addressed and a three‐level simple and effective checking method is developed. A case study is presented with a hyperspectral data set recorded by the Pushbroom Hyperspectral Imager (PHI) and related GIS data.     

Parallel fine spatial resolution satellite sensor image segmentation based on an improved Pulse‐Coupled Neural Network

An improved Pulse‐Coupled Neural Network (PCNN), which has two modules with different inner processing mechanisms, was introduced and applied to segment an IKONOS image with different parameters. Results show that the improved PCNN can segment the image in a parallel and multiscale way, and the output image is easier to analyse than with the basic PCNN.     

Unsupervised segmentation of medical image based on difference of mutual information

Image segmentation is a very important research field in the scope of image process- ing. It has extensive application and involves almost all fields such as image understand- ing, pattern recognition and image encoding, etc. Furthermore, research of imag…     

SVM‐based segmentation and classification of remotely sensed data

Support Vector Machines (SVM) is becoming a popular alternative to traditional image classification methods because it makes possible accurate classification from small training samples. Nevertheless, concerns regarding SVM parameterization and computational effort have arisen. This Letter is an evaluation of an automated SVM‐based method for image classification. The method is applied to a land‐cover classification experiment using a hyperspectral dataset. The results suggest that SVM can be parameterized to obtain accurate results while being computationally efficient. However, automation of parameter tuning does not solve all SVM problems. Interestingly, the method produces fuzzy image‐regions whose contextual properties may be potentially useful for improving the image classification process.     

Parameter selection for region‐growing image segmentation algorithms using spatial autocorrelation

Region‐growing segmentation algorithms are useful for remote sensing image segmentation. These algorithms need the user to supply control parameters, which control the quality of the resulting segmentation. An objective function is proposed for selecting suitable parameters for region‐growing algorithms to ensure best quality results. It considers that a segmentation has two desirable properties: each of the resulting segments should be internally homogeneous and should be distinguishable from its neighbourhood. The measure combines a spatial autocorrelation indicator that detects separability between regions and a variance indicator that expresses the overall homogeneity of the regions.     

Comparison and improvement of wavelet‐based image fusion

The wavelets used in image fusion can be categorized into three general classes: orthogonal, biorthogonal, and non‐orthogonal. Although these wavelets share some common properties, each wavelet also has a unique image decomposition and reconstruction characteristic that leads to different fusion results. This paper focuses on the comparison of the image‐fusion methods that utilize the wavelet of the above three general classes, and theoretically analyses the factors that lead to different fusion results. Normally, when a wavelet transformation alone is used for image fusion, the fusion result is not good. However, if a wavelet transform and a traditional fusion method, such as an IHS transform or a PCA transform, are integrated, better fusion results may be achieved. Therefore, this paper also discusses methods to improve wavelet‐based fusion by integrating an IHS or a PCA transform. As the substitution in the IHS transform or the PCA transform is limited to only one component, the integration of the wavelet transform with the IHS or PCA to improve or modify the component, and the use of IHS or PCA transform to fuse the image, can make the fusion process simpler and faster. This integration can also better preserve colour information. IKONOS and QuickBird image data are used to evaluate the seven kinds of wavelet fusion methods (orthogonal wavelet fusion with decimation, orthogonal wavelet fusion without decimation, biorthogonal wavelet fusion with decimation, biorthogonal wavelet fusion without decimation, wavelet fusion based on the ‘à trous’, wavelet and IHS transformation integration, and wavelet and PCA transformation integration). The fusion results are compared graphically, visually, and statistically, and show that wavelet‐integrated methods can improve the fusion result, reduce the ringing or aliasing effects to some extent, and make the whole image smoother. Comparisons of the final results also show that the final result is affected by the type of wavelets (orthogonal, biorthogonal, and non‐orthogonal), decimation or undecimation, and wavelet‐decomposition levels.     

Kapur’s entropy based optimal multilevel image segmentation using Crow Search Algorithm

Image segmentation is an essential part of image analysis, which has a direct impact on the quality of image analysis results. Thresholding is one of the simplest and widely used methods for image segmentation. Thresholding can be either bi-level, which involves partitioning of an image into two segments, or multilevel, which partitions an image into multiple segments using multiple thresholds values. This paper focuses on multilevel thresholding. A good segmentation scheme through multilevel thresholding identifies suitable threshold values to optimize between-class variance or entropy criterion. For such optimizations, nature inspired metaheuristic algorithms are commonly used. This paper presents a Kapur’s entropy based Crow Search Algorithm (CSA) to estimate optimal values of multilevel thresholds. Crow Search Algorithm is based on the intelligent behavior of crow flock. Crow Search Algorithm have shown better results because of less number of parameters, no premature convergence, and better exploration–exploitation balance in the search strategy. Kapur’s entropy is used as an objective function during the optimization process. The experiments have been performed on benchmarked images for different threshold values (i.e. 2, 4, 8, 16, 32 thresholds). The proposed method has been assessed and performance is compared with well-known metaheuristic optimization methods like Particle Swarm Optimization (PSO), Differential Evolution (DE), Grey Wolf Optimizer (GWO), Moth-Flame Optimization (MFO) and Cuckoo Search (CS). Experimental results have been evaluated qualitatively and quantitatively by using well-performed evaluation methods namely PSNR, SSIM, and FSIM. Computational time and Wilcoxon p-type value also compared. Experimental results show that proposed algorithm performed better than PSO, DE, GWO, MFO and CS in terms of quality and consistency.     

Gesture image segmentation with Otsu’s method based on noise adaptive angle threshold

By analyzing the essence and deficiency of the improved Otsu’s method, this paper proposes a noise adaptive angle threshold based Otsu’s method for gesture image segmentation. It first designs a two-dimensional histogram of gray value-neighborhood truncated gray mean to avoid the interference of extreme noise by discarding the extremes of the neighborhood. Then, the probability that the pixel is noise is calculated according to the actual situation, adaptive filtering is implemented to enhance the algorithm’s universal applicability. It finally converts the threshold space to an angle space from 0° to 90°, and the threshold search range is compressed to improve its efficiency. As the gesture is close to the background and the boundary is blurred, this paper combines the global and local Otsu’s method to segment the gesture images based on the angle space. On the one hand, it uses the global Otsu’s method to obtain the global threshold t₁. On the other hand, it uses the local Otsu’s method to obtain the local threshold t₂, and segments gesture images based on t₂. Experimental results show that the proposed method is effective and can accurately segment gesture images with different noises.

     

Grayscale image segmentation by spatially variant mixture model with student’s t-distribution

A spatially variant finite mixture model with Student’s t-distribution component function is proposed for grayscale image segmentation. This model employs a new weight function which contains the information along the different spatial directions indicating the relationship of the pixels in the neighborhood. The label probability proportions are explicitly represented as probability vectors in the model. Gradient descend method is used to update the unknown parameters. The proposed model contains fewer parameters and it is easy to be implemented compare with the Markov random field (MRF) models. Comprehensive experiments on synthetic and natural images are carried out to demonstrate that the proposed model outperforms some other related ones.     

Transreceiving of encrypted medical image – a cognitive approach

Multimedia Tools and Applications - Recently, there is an increasing demand for efficient and secure transreception of medical images in telemedicine applications. Though a fixed spectrum is...     

Satellite evidence of a topographic sink‐source system in a small semi‐arid watershed

Water is a limiting factor for biological processes in drylands and consequently it is expected that vegetation cover along slopes will be affected by runoff flow regime and shape of topography. Nevertheless, spaceborne remote sensing data have hardly been used to study the effect of hydrological processes on vegetation pattern at the slope scale. This Letter reports on the spatio‐temporal variation in spaceborne‐derived Normalized Difference Vegetation Index (NDVI) data, in physiographic units of unique pedo‐hydrological characteristics, in a semi‐arid watershed. It was found that NDVI values in the footslope and shoulder physiographic units are significantly (p<0.0001) higher than those of the interfluve and the backslope. This difference, observed during the entire phenological cycle, was enhanced towards the peak season but was of less significance during the early season and towards senescence. These results support the hypothesis that water redistribution can significantly increase plant production in sink areas, also in the slope‐scale.     

A multi‐scale segmentation approach to filling gaps in Landsat ETM+ SLC‐off images

On 31 May 2003, the Landsat Enhanced Thematic Plus (ETM+) Scan Line Corrector (SLC) failed, causing the scanning pattern to exhibit wedge‐shaped scan‐to‐scan gaps. We developed a method that uses coincident spectral data to fill the image gaps. This method uses a multi‐scale segment model, derived from a previous Landsat SLC‐on image (image acquired prior to the SLC failure), to guide the spectral interpolation across the gaps in SLC‐off images (images acquired after the SLC failure). This paper describes the process used to generate the segment model, provides details of the gap‐fill algorithm used in deriving the segment‐based gap‐fill product, and presents the results of the gap‐fill process applied to grassland, cropland, and forest landscapes. Our results indicate this product will be useful for a wide variety of applications, including regional‐scale studies, general land cover mapping (e.g. forest, urban, and grass), crop‐specific mapping and monitoring, and visual assessments. Applications that need to be cautious when using pixels in the gap areas include any applications that require per‐pixel accuracy, such as urban characterization or impervious surface mapping, applications that use texture to characterize landscape features, and applications that require accurate measurements of small or narrow landscape features such as roads, farmsteads, and riparian areas.     

An image segmentation method based on Mumford–Shah model with mask factor and neighborhood factor

Pattern Analysis and Applications - A novel image segmentation model is proposed to improve the stability of existing segmentation methods. In the proposed model, we introduce two factors into the...     

Phase‐correlation‐based hyperspectral image classification using multiple class representatives obtained with k‐means clustering

In this letter, a modification to a phase–correlation‐(PC‐)based supervised classification method for hyperspectral data is proposed. An adaptive approach using different numbers of multiple class representatives (CRs) extracted using PC‐based k‐means clustering for each class is compared with the use of selecting a small, pre‐determined number of dissimilar CRs. PC is used as a distance measure in k‐means clustering to determine the spectral similarity between each pixel and cluster centre. The number of representatives for each class is chosen adaptively, depending on the number of training samples in each class. Classification is performed for each pixel according to the maximum value of PCs obtained between test samples and the CRs. Experimental results show that the adaptive method gave the highest classification accuracy (CA). Experiments on the effect of reducing the size of the feature vectors found that CA increased as the feature vector decreased.     

Incorporating domain knowledge and spatial relationships into land cover classifications: a rule‐based approach

For some tropical regions, remote sensing of land cover yields unacceptable results, particularly as the number of land cover classes increases. This research explores the utility of incorporating domain knowledge and multiple algorithms into land cover classifications via a rule‐based algorithm for a series of satellite images. The proposed technique integrates the fundamental, knowledge‐based interpretation elements of remote sensing without sacrificing the ease and consistency of automated, algorithm‐based processing. Compared with results from a traditional maximum likelihood algorithm, classification accuracy was improved substantially for each of the six land cover classes and all three years in the image series. Use of domain knowledge proved effective in accurately classifying problematic tropical land covers, such as tropical deciduous forest and seasonal wetlands. Results also suggest that ancillary data may be most useful in the classification of historic images, where the greatest improvement was observed relative to results from maximum likelihood. The cost of incorporating contextual knowledge and extensive spatial data sets may be justified, since results from the proposed technique suggest a considerable improvement in accuracy may be achieved.