The use of face recognition based on numerous sketches is crucial for both law enforcement and digital entertainment. Due to lack of information given by the victim, it is difficult to identify faces from several sketches. In this study, we provide an unique SIFT (Scale Invariant Feature Transformation), 6-point facial landmark, Intuitionistic Fuzzy (IF) and Fuzzy \(\:{m}_{X}^{*}\) oscillation based techniques. To improve accuracy of face recognition, we have used fuzzy based similarity measuring techniques called IF and Fuzzy \(\:{m}_{X}^{*}\) oscillation. First, apply SIFT on sketches and digital pictures to find keypoints, then select the keypoints for extracting features. These values are classified using IF and fuzzy \(\:{m}_{X}^{*}\) oscillation after feature extraction. Since we can get more than one image as output, so to overcome this problem we use a 6-point facial landmark detector (which is a modified facial landmark detector of 68-point facial landmark detector) for both the sketch and resulting digital images. Using these 6 points we draw two different regions for further verification. Two possibilities, which are presented at the beginning of Sect. 6, were taken into consideration for the experiment. As a result, by utilizing the provided method, matching a facial sketch to a photo will make it straightforward to identify the correct image. Experimental results shows that our approach achieved 95.1% accuracy for Fuzzy \(\:{m}_{X}^{*}\) oscillation with SIFT domain and 97.3% for IF \(\:{m}_{X}^{*}\) oscillation with SIFT domain. Our algorithm’s accuracy suggested that IF and Fuzzy \(\:{m}_{X}^{*}\) oscillation may be applied to the field of facial recognition-based sketching.
