Face Recognition Based on Windowing Techniques, with Compressed Hybrid Domain Features
Niranjana Kumara M., Abbai Reddy Divya, K. Bommanna Raja, K R Venugopal · International Journal of Electronics and Communication Engineering · 2023
Face recognition is used in human-machine interfaces for day-to-day activities in real-time applications. This paper proposes face recognition based on windowing techniques with compressed hybrid domain features. The face images are resized to 224×160 and are segmented into windows of sizes 4 × 4, 8 × 8, and 16 × 16. The hybrid features are extracted using Discrete Wavelet Transform (DWT) and the covariance concept on each window. The average covariance of each window is computed to obtain compressed final features. Artificial Neural Network (ANN) is used for classification to identify a person effectively. The obtained results of average recognition are around 98% for four benchmarked public face databases. It is observed that the proposed model attains a higher recognition rate with a reduced number of features compared with the existing methods.