vioft2nntf2t|tblJournal|Abstract_paper|0xf4ff9ab42b000000150e000001000800
This article introduces a face expression using a processing sequence that involves the CNN 3D Skull Description ROI point extraction. The first step is the noise reduction process using gaussian filtering from the input 3D face image. The functionality is extracted with the Otsu Thresholding Multilevel Operation. Using morphological erosion, boring off, and region operators, the derived features from the photos are improved. The extraction of points is then sent to the Interest Area (ROI) which essentially extracts points in both photos. The experiment is performed on a sequence of 100 photographs, including the cranial and facial pictures. The extracted ROI of the enriched input image is sent to the CNN, which is training and testing new input images directly applied to the CNN classification. The trained ROIs are transmitted as trained data. For 3D modelling, the CNN classification is used and finally, face expression is used to compare the similitude between the original image and the image being tested. The result shows that a higher classification rate for the proposed method is achieved, efficiently extracting the points and improving the classification required for 3D modelling than existing methods.