The landmark variation improvement on the different modalities for the facial sketch features detection

Facial feature detection studies on the same modality have been conducted by many researchers, but the research results cannot be implemented on the different modality, only a few studies that can be used to detect the facial features on the different modality. In this research, we proposed method to detect the facial features on the different modality. The deviation standard on the landmark variations improvement has been considered as parameters to improve the moving direction toward the corresponding features. The experimental results show that the detection accuracy of our proposed method is 91.944% for the 1^st model and 91.46% for the 2^nd model. Our proposed method has been shown outperformed the mixture model method.