echemi logo
  • Product
  • Supplier
  • Inquiry
    Home > Chemicals Industry > China Chemical > Stacked spectral feature space maps: CNN-based crop classification of hyperspectral remote sensing images

    Stacked spectral feature space maps: CNN-based crop classification of hyperspectral remote sensing images

    • Last Update: 2022-05-02
    • Source: Internet
    • Author: User
    Search more information of high quality chemicals, good prices and reliable suppliers, visit
      [ Hot Focus on Chemical Machinery and Equipment Network ] People take food as their heaven
    The importance of food in human daily life is irreplaceable

    And the production of food, whether it is a processed product or a primary product, is inseparable from crops

    Crops include food crops, cash crops (such as oil crops, vegetable crops, etc.
    ) two categories


    Chemical machinery and equipment network hotspots focus on chemical machinery and equipment
      Identifying and classifying crop varieties is conducive to strengthening the management and rational promotion of new crop varieties
    Hyperspectral remote sensing data contains rich spectral information, which is widely used in the monitoring of crop distribution and dynamic changes, and plays an irreplaceable role in the classification of crop types in precision agriculture

    At present, the utilization of features mainly includes traditional feature selection involving expert knowledge and automatic feature selection closely combined with Convolutional Neural Networks (CNN)

    CNN automatic feature selection can automatically extract high-level domain-oriented features from the input data to achieve higher classification accuracy

    But compared with mining spatial features, CNN is still insufficient in mining spectral features



      At present, traditional feature selection has been proved by related research to improve the accuracy of multi-class classifiers including CNN.
    Combining traditional feature selection methods with CNN advanced spatial feature automatic extraction is a gradually popular classification strategy

    However, the current combined methods do not comprehensively utilize spatial features and spectral information, and do not reflect the rich spectral information of hyperspectral images in crop classification

    At the same time, using traditional feature selection to mine spectral features and combining with CNN to automatically extract domain-oriented high-level features remains to be further studied



      Recently, Beijing Normal University and the Institute of Crop Science of the Chinese Academy of Agricultural Sciences published a research paper online in The Crop Journal, proposing a novel spectral feature—stacked spectral feature space patch.
    SSFSP) for CNN-based crop classification of hyperspectral remote sensing images

    This feature converts original recessive spectral features into dominant spatial features, which can be combined with 2D CNN to mine spectral and spatial features simultaneously

    A comparative study of multiple high-spatial-resolution hyperspectral datasets shows that SSFSP can achieve higher classification accuracy than raw spectral inputs



      Original title: Stacked spectral feature space maps: CNN-based crop classification of hyperspectral remote sensing images
    This article is an English version of an article which is originally in the Chinese language on and is provided for information purposes only. This website makes no representation or warranty of any kind, either expressed or implied, as to the accuracy, completeness ownership or reliability of the article or any translations thereof. If you have any concerns or complaints relating to the article, please send an email, providing a detailed description of the concern or complaint, to A staff member will contact you within 5 working days. Once verified, infringing content will be removed immediately.

    Contact Us

    The source of this page with content of products and services is from Internet, which doesn't represent ECHEMI's opinion. If you have any queries, please write to It will be replied within 5 days.

    Moreover, if you find any instances of plagiarism from the page, please send email to with relevant evidence.