Figure 3
From: VNIR–NIR hyperspectral imaging fusion targeting intraoperative brain cancer detection
Segmentation and classification maps of three examples of the test set from the HSI plastic dataset and average Jaccard and accuracy results obtained from the thirteen images. Color, material, and material-color segmentation (a–c) and classification (d–f) problems, respectively, using VNIR, NIR, and fused data. Each column (from left to right) represents the RGB images obtained with a digital camera, the ground-truth maps, the VNIR, NIR, and fused segmentation results, respectively, and the average Jaccard and accuracy results obtained with the entire test set for the three different segmentation and classification algorithms. Results were statistically analyzed using a paired, one-tailed Student’s t test at the 5% significance level. (*) Statistically significant difference (\(p < 0.05\)). (**) Highly statistically significant difference (\(p < 0.001)\). ABS: Acrylonitrile Butadiene Styrene; PLA: Polylactic Acid; PETG: Polyethylene Terephthalate Glycol. HKM: Hierarchical K-means; SVM: Support Vector Machines; RF: Random Forest; KNN: K-Nearest Neighbors.
