Improving Single Image-based Face Morphing Attack Detection Rate with Image Pre-processing

Morphing attack detection is nothing but person who can cross the border without any valid identity. Face morphing attack detection can be divided into two categories: S-MAD for Single Morph Attack Detection and D-MAD for Differential Morph Attack Detection. The proposed study in this research is mainly focused on pre-processing methods using various filtering approaches. Several filtering approaches are used to decrease noise from facial images in order to distinguish changing faces. In this paper, the non-local means filter, the gaussian filter, the low-pass filter, and a bilateral filter are analysed and compared. The performance and quality of Feature Extraction, Optimization, and Classification improves when the bilateral filter technique is applied for image preprocessing. The AMSL Face Morph picture dataset is collected via the Github Repository, and an experiment is carried out to evaluate the efficiency of bilateral filtering in removing noise from the image.