An Optimal Texture Pattern Model Of Face Recognition Using MMOO-CDNN

Different forms of data prediction models are used in image processing to recognize the type of data and predict its class label. Among these models, the texture pattern-based occluded face recognition method stands out as it improves the prediction performance compared to other feature representation methods. However, this method also increases the size of features for the images, resulting in the need for big data processing. The proposed model focuses on recognizing the form of occlusion type and identify the person by the occluded face image. To conduct this work, we propose the texture pattern for extracting image using the Flit Synchronized Texture Pattern (FSTP), which implements a distribution function to represent the image texture pattern based on the intensity of the image matrix. Furthermore, the Multi-Model Osprey Optimization (MMOO) method is employed to optimize the texture feature for the big data of image texture feature matrix. This ensures that the image classification model is efficient and effective in handling large amounts of data. To classify the face and to identify the person from occluded face images, Correlative Directional Neural Network model is utilized and the performance of the proposed model is validated using Real World Occluded Faces (ROF) dataset consisting of 5559 images of 180 persons. The results and comparative analysis demonstrate that the proposed model outperforms traditional classification and prediction methods in terms of performance enhancement.