Face Recognition Image Processing Technology Based on SIFT Algorithm
Abstract
With the development of image recognition technology, the processing of human face information can confirm human identity. Facial recognition technology has been widely used in many fields, effectively improving the efficiency of facial information entry and recognition. The core of the face recognition process is an analysis of the characteristics of face images. The traditional facial image recognition method involves a local binary pattern (LBP) algorithm, which has high recognition accuracy when facial images contain complete information. However, actual collected face images can be affected by various environmental factors, and traditional image recognition methods find it difficult to accurately determine facial characteristics. This paper applied a scale invariant feature transform (SIFT) algorithm to facial image recognition, and compared and analyzed the traditional LBP algorithm and SIFT algorithm in respect to four factors: illumination intensity, facial expression, image occlusion ratio, and face offset angle. Experimental results showed that for male facial images, the average face recognition accuracy rates of the LBP algorithm and the SIFT algorithm under different light intensities were 94.64% and 99.52%, respectively. For female facial images, the average face recognition accuracy rates of the LBP algorithm and the SIFT algorithm under different light intensities were 92.04% and 99.08%, respectively. Therefore, the application of SIFT algorithms can improve the accuracy of facial recognition under different light intensities.
Keywords: Image processing; facial recognition; scale invariant feature transform (SIFT); local binary patterns
Cite As
H. Zhao, P. Li, "Face Recognition Image Processing Technology Based on SIFT Algorithm",
Engineering Intelligent Systems, vol. 32 no. 5, pp. 477-485.