“Face Retrieval System by Similarity of Impression Based on Hair Attribute” by Fuji, Fukusato, Sasaki, Masuda, Hirai, et al. …

  • ©Takahiro Fuji, Tsukasa Fukusato, Shoto Sasaki, Taro Masuda, Tatsunori Hirai, and Shigeo Morishima

Conference:


Type:


Entry Number: 45

Title:

    Face Retrieval System by Similarity of Impression Based on Hair Attribute

Presenter(s)/Author(s):



Abstract:


    There is a great demand for human image retrieval on the Web. On searching person’s images, not only whether he or she is the person you are looking for, but also the appearance of the person is an important factor. Since current image search systems such as Google Image Search are based on textual data, they are useful when we want to look over a particular person’s face, but they cannot consider the appearance or the impression of the person sufficiently. Most of the images on the Web are not annotated because it is often difficult to describe the human appearances in words.
    While there are many factors, hair is especially an important feature of human appearance. According to Davis et al. [1981], hair appearance is the most important feature in the recognition of familiar people. Therefore, in this paper we present a human image retrieval system based on hair appearance features. We propose novel hair appearance features which affects the impression of the person.

References:


    1. Davis, G., Ellis, H., and Shepherd, J., eds. 1981. Perceiving and Remembering Faces, London: Academic Press, Academic Press series in cognition and perception.

PDF:



ACM Digital Library Publication:



Overview Page: