In the field of computer vision and pattern recognition, face orientation recognition stands as a significant topic. In the paper A Robust Approach of Facial Orientation Recognition from Facial Features, the authors Stefan Andrei, Kishor Datta Gupta, Md Manjurul Ahsan and Kazi Md. Rokibul Alam introduce us to an image mapping technique for face analysis.
The methodology of the study consists of two main phases: Face Feature Extraction and creating the graph image, and Matching Graph image with stored images. The Face Feature Extraction presents four steps that include Face Detection, Feature Extraction, Obtaining Feature Data and Creating image with these data.
The first step in facial feature detection is detecting the face. This requires analyzing the entire image. The second step is using the isolated face(s) to detect each feature. The technique of this study relies on the four main features of the face: left eye, right eye, nose, and mouth. It is mandatory to obtain the positions, size, height, width, and angle of these features respective to each face. By acquiring the data from the features, a new picture including the model and shape of the face is created.
The phase Matching Graph image calculates two images from all pixels, and calculates the distance between the two images’ pixels. If the images are close to the stored image, it passes as recognized.
After image matching, the authors obtained 93% positive results, for 1000 random sample images tested on the nine criteria of orientation.
The authors’ method could be utilized for artificial intelligence, game controller development, as well as traffic control and robot development. This improved method proves to be faster, requires less stored data and can work faster in short periods of time.