An Improved Image Recognition Technique Based on SPCA-GA-SVM: Dimensionality Reduction and Classification

Image recognition technology is based on the main features of the image, using computers to process and analyze the image. Recognizing various targets in different modes can perform a series of enhancements and reconstructions on the image to improve image quality, identify relevant information, etc. Researchers have developed different machine learning methods for image recognition. Among them, dimensionality and prediction accuracy are important issues that have always existed in recognition. Researchers improve prediction accuracy by optimizing dimensionality reduction and classification algorithms. Therefore, this study presents a hybrid model that integrates three algorithms, namely Sparse Principal Component Analysis (SPCA) for dimensionality reduction, Support Vector Machine (SVM) for classification, and Genetic Algorithm (GA) for SVM optimization. The model was optimized for dimensionality reduction and classification algorithms respectively, and the optimized algorithms were combined to construct a hybrid model: SPCA-GA-SVM. Applying this model to the LFW face image dataset, the results obtained indicate that the recognition accuracy is improved. The accuracy of the method was 89.70% and the test MSE was 0.734. © 2024 Copyright held by the owner/author(s).