facial expressions are the most powerful and direct means of presenting human emotions and feelings and offer a window into a persons’ state of mind. in recent years, the study of facial expression and recognition has gained prominence; as industry and services are keen on expanding on the potential advantages of facial recognition technology. as machine vision and artificial intelligence advan...

many researchers adopt local binary pattern for pattern analysis. however, the long histogram created by local binary pattern is not suitable for large-scale facial database. this paper presents a simple facial pattern descriptor for facial expression recognition. local pattern is computed based on local gradient flow from one side to another side through the center pixel in a 3x3 pixels region...

Automatic facial recognition has many potential applications in different areas of humancomputer interaction. However, they are not yet fully realized due to the lack of an effectivefacial feature descriptor. In this paper, we present a new appearance based feature descriptor,the local directional pattern (LDP), to represent facial geometry and analyze its performance inrecognition. An LDP feat...

automatic facial recognition has many potential applications in different areas of humancomputer interaction. however, they are not yet fully realized due to the lack of an effectivefacial feature descriptor. in this paper, we present a new appearance based feature descriptor,the local directional pattern (ldp), to represent facial geometry and analyze its performance inrecognition. an ldp feat...

This paper presents a novel method for interest region description. We adopted the idea that the appearance of an interest region can be well characterized by the distribution of its local features. The most well-known descriptor built on this idea is the SIFT descriptor that uses gradient as the local feature. Thus far, existing texture features are not widely utilized in the context of region...

Many researchers adopt Local Binary Pattern for pattern analysis. However, the long histogram created by Local Binary Pattern is not suitable for large-scale facial database. This paper presents a simple facial pattern descriptor for facial expression recognition. Local pattern is computed based on local gradient flow from one side to another side through the center pixel in a 3x3 pixels region...