387 method. The labelled images are stored as a training data set and used in the subsequent model generation process. (2) Discriminant Model Generation Upon completion of the step, the user is then able to initiate the building of a discriminant model by clicking the "Generate Model" button, as illustrated in the Figure 6(c). During this process, the status on the application changes to "Generating" and "Generated" is displayed upon completion of the model construction. The model can be generated again using the same procedure to create a discrimination model that reflects the latest information, should new images be added to the data set or existing labels be modified. (3) Application of the decision model to unknown images, the discriminant model that has been generated can also be used to classify unknown hyperspectral images. The user uploads the image to be judged to the application and applies the discriminant model to it. The application employs the model to assign each pixel in the uploaded image to its designated label. The result of this process is then displayed visually by means of color coding according to each label. For example, the label "Component A" is mapped in red, "Component B" in blue, and "Others" in green, as illustrated in Figure 2(d). This visual representation facilitates intuitive comprehension of the composition and distribution of components across the entire image. Figure 6 Workflow of the application: (a) Uploading hyperspectral image; (b) Labeling selected regions; (c) Generating a discriminant model; (d) Applying the model to an unknown image with visualized results.
RkJQdWJsaXNoZXIy MTM0Mzk2