Proposed Dense Block for Reconstruction Hyperspectral Imaging of Milk Object

One of the techniques in image acquisition based on spatial information in the near infrared visible spectrum range is only hyperspectral imaging. Based on the advantages of hyperspectral, it can be seen that the use of hyperspectral is only among professionals with very expensive operational costs and complicated measurement designs, so a reconstruction is carried out using RGB imagery. This study aims to reconstruct RGB images from Hyperspectral images using objects from milk using the Proposed Algorithm Model Dense Block. RGB resulting from conversion using CIE 1931 Color Matching Function. The dataset used includes Training, Valdiation, and Testing with different target bands and spatial images to determine the best performance of the Reconstruction Algorithm. The test results of the performance model use an evaluation technique using the RMSE and MAE metrics. Band variations and segmentation can affect the performance of the Reconstruction Algorithm. The value of each RMSE and MAE is 0.01 and 0.023 from the value it can be seen that the reconstruction can be done with the proposed dense block.