Hyperspectral image feature selection for the fuzzy c-means spatial and spectral clustering

Salem, Manel Ben; Ettabaa, Karim Saheb; Bouhlel, Med Salim

doi:10.1109/ipas.2016.7880114

Cited by 12 publications

(5 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…FCM assigns the "e" proportion of relevance to "e" classes for each pixel in the image. This feature allows reconsideration of pixel membership to image classes in further iterations of the algorithm and illustrates more appropriately the natural aspect of uncertainty existing in the data [59]. This helps identify the target pixels of endmembers to be extracted.…”

Section: Figure 1 Block Diagram Of the Proposed Methodologymentioning

confidence: 98%

Endmember Estimation Using Fuzzy Grade of Membership and Spectral Matching

Devi,

Kalaivani

2023

IEEE Access

View full text Add to dashboard Cite

Spectral unmixing in hyperspectral images involves determining endmembers and their associated abundance maps. The endmember estimate is extremely important in the processing of high resolution hyperspectral data. This study provides a unique automatic method for extracting endmembers by integrating fuzzy clustering and a spectral-matching approach. The number of endmembers in an image is estimated in the first stage using the Hysime algorithm. Data are subsequently classified using a fuzzy c-means algorithm, which determines the grade of membership values for each data point. A threshold operation on membership values is used to select a collection of pixels as target pixels. Spectral matching aids in the selection of target pixels and searches for a specific endmember within a cluster. Endmember bundles are extracted from target pixels and compared with ground truth data using a spectral angle mapper. The performance of the proposed technique was evaluated in two ways: directly on full hyperspectral data and on dimension-reduced data by employing one simulated and two real datasets. The mean spectral angle and root mean square error were used as performance indicators. Furthermore, the accuracy of extracted endmembers was validated by creating abundance maps using a fully constrained least square technique, and the results were analyzed.

show abstract

Section: Figure 1 Block Diagram Of the Proposed Methodologymentioning

confidence: 98%

Endmember Estimation Using Fuzzy Grade of Membership and Spectral Matching

Devi,

Kalaivani

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…In [17], devoted to the selection of hyperspectral image characteristics for spatial and spectral clustering by the fuzzy C-means logic method, the Ward's method is used as an agglomerative algorithm for constructing a hierarchy in which each spectral band is considered as a cluster. The iterative merging process is repeated until the desired number of clusters is reached.…”

Section: Overview and Comparisonmentioning

confidence: 99%

The Application of the Model of High-Speed Pixel Clustering in Problems of Preprocessing of the Images of the Remote Sensing of the Earth

Khanykov

2020

Proceedings of the 30th International Conference on Computer Graphics and Machine Vision (GraphiCon 2020). Part 2

View full text Add to dashboard Cite

The purpose of the research is to use the modified Ward’s method in high-speed processing of full-HD images of the remote sensing of the Earth. The classical Ward’s method is modified by dividing the computational process into three successive stages. The first stage quickly builds a coarse hierarchy of approximations. The second stage performs a quality improvement of the specified partition for a fixed number of colors (clusters). The third stage is the clustering of the superpixels using the Ward’s method. The software-algorithmic toolkit consists of four operations on clusters of pixels and image segments: merge operation joins together two clusters; divide operation reversibly disjoins the selected cluster into two; split operation extracts the part of the cluster into individual cluster; correct operation reclassifies pixels by extracting from one cluster and inserting into another cluster. The quality is assessed by the total squared error. The quality improvement is provided by iterative execution of a combination of merge and divide operations of pixel clusters, in particular image segments. One of the clusters (segments) is divided in two and a pair of other mismatched with it is combined into one according to the criterion of the minimum increment of the total squared error. The proposed modified Ward’s method is appropriate in processing of fullHD images of the remote sensing of the Earth. The results of processing in pure segmentation and clustering modes are compared. The proposed pixel clustering model is appropriate in high-speed processing of the fullHD images. The pixel clustering in comparison with image segmentation allows to define in more detail both the contours of objects of interest and their internal structure

show abstract

“…In the work of Liu et al [22], spatial structural attributes and statistical information is combined and represented into the positive matrix factorization to handle dynamicity associated with the unmixing process. Salem et al [23] suggested a band selection mechanism oriented on the layer-based spectral and spectrum information using C-means clustering. The work of Wo et al [24] introduced a framework to assess different techniques from different perspectives, such as feature representation and classification performance.…”

Section: Introductionmentioning

confidence: 99%

Precise identification of objects in a hyperspectral image by characterizing the distribution of pure signatures

Panchal¹,

Shivaputra²

2022

IJECE

View full text Add to dashboard Cite

<span lang="EN-US">Hyperspectral image (HSI) has been widely adopted in many real-world applications due to its potential to provide detailed information from spectral and spatial data in each pixel. However, precise classification of an object from HSI is challenging due to complex and highly correlated features that exhibit a nonlinear relationship between the acquired spectral unique to the HSI object. In literature, many research works have been conducted to address this problem. However, the problem of processing high-dimensional data and achieving the best resolution factor for any set of regions remains to be evolved with a suitable strategy. Therefore, the proposed study introduces simplified modeling of the hyperspectral image in which precise detection of regions is carried out based on the characterization of pure signatures based on the estimation of the maximum pixel mixing ratio. Moreover, the proposed system emphasizes the pixel unmixing problem, where input data is processed concerning wavelength computation, feature extraction, and hypercube construction. Further, a non-iterative matrix-based operation with a linear square method is performed to classify the region from the input hyperspectral image. The simulation outcome exhibits efficient and precise object classification is achieved by the proposed system in terms classified HSI object and processing time.</span>

show abstract

Hyperspectral image feature selection for the fuzzy c-means spatial and spectral clustering

Cited by 12 publications

References 12 publications

Endmember Estimation Using Fuzzy Grade of Membership and Spectral Matching

Endmember Estimation Using Fuzzy Grade of Membership and Spectral Matching

The Application of the Model of High-Speed Pixel Clustering in Problems of Preprocessing of the Images of the Remote Sensing of the Earth

Precise identification of objects in a hyperspectral image by characterizing the distribution of pure signatures

Contact Info

Product

Resources

About