Article Preview
Top1. Introduction
Image segmentation is one of the intermediate level computer vision process, which performs the jointly grouping of image regions into distinct parts of objects. From an implementation point of view, it is the primary task in computer vision, which allows the computer to understand and see the image contents, classify a region or pixel in the image, and divide the image into different parts according to visual understanding. Face segmentation is a basic task in face image analysis. In face segmentation, a computer-based algorithm segments a face image according to the different regions in the face. Semantic face segmentation allows the computer to understand the face image contents at the pixel level. As such, for semantic face segmentation, a number of complex features are also employed (K. Khan, R. U. Khan, K. Ahmad, F. Ali and K. Kwak, 2020).
In present decades, human face image processing is extensively employed in dissimilar fields with the increasing demands of commercial applications (He, H., 2020)(Veerashetty, S., Patil, N.B., 2020) (S.Shanmugapriya, B.S.Sathishkumar,2014) (Virupakshappa, Basavaraj, A, 2019) (Ambika, Rajkumar L. Biradar & Vishwanath Burkpalli, 2019)and security(Kavitha G.,, & Elango N. M., 2020)(Virupakshappa, Amarapur B., 2018) (Onuchowska, A., & de Vreede, G., 2017)(Ambika, Biradar, R.L.,2020) (SachinKumar Veerashetty, 2021)(Y. Zhao, F. Tang, W. Dong, F. Huang, and X. Zhang,2019). Face analysis systems are regularly utilized in numerous applications such as style transfer, face swap, virtual makeup, emotion recognition, etc (M. Ferrara, A.Franco, and D. Maio, 2012). Recently, many techniques are developed for perceiving the multimodal information to understand and comprehension of face images and videos, which is considered as an important aspect in multimedia and computer vision applications (X.Ding, Y. Luo, L.Sun, and F. Chen, 2014)(I. Ahn, and C. Kim, 2016).
In recent period, several methods are developed by the researchers for face segmentation such as viola jones algorithm (A.Hernandez-Matamoros, A.Bonarini, E. Escamilla-Hernandez,M. Nakano-Miyatake, and H. Perez-Meana, 2016)(A.Setu, and M. Rahman, 2016), super pixel based segmentation (K. Luu, C. Zhu, C. Bhagavatula,T.H.N. Le, and M.Savvides, 2016)(K. Khan, N. Ahmad, K.Ullah, and I.Din, 2017), adaptive threshold based segmentation(M.C.Sanchez-Cuevas, R.M.Aguilar-Ponce, and J.L. Tecpanecatl-Xihuitl, 2013), etc. Tremendous growth is attained in facial investigation tasks like face detection, tracking, and face attributes recognition by considering the benefit of recent developments in machine learning. In earlier research works, researchers used different segmentation algorithms for segmenting the facial parts (back-ground, skin, nose, hair, mouth, and eyes).A few major issues in the existing studies are: poor sharpening of edges, highly sensitive to noise, different facial expressions, and illumination conditions, etc. In order to address these concerns, a new segmentation algorithm is proposed for reliable and accurate segmentation of facial parts.