ชื่อผลงานทางวิชาการ
Affective word indexing for Image Retrieval Based on Color and Emotion relationship
ชื่อนักวิจัย
หมวดหมู่ผลงาน
ประเภทผลงาน
บทคัดย่อ
An automatic method for image indexing that uses affective
words which can be associated with quantitative emotion-color features is
described. The features are derived from color-emotion relationship models, and
art-inspired color features. Color attributes such as hue, brightness,
saturation, chromaticity, and lightness are considered. The similarities of
affective words and image contents are analyzed in the emotion-color feature
space. Consequently, the degree of similarity of the affective words and images
can be measured explicitly in terms of a quantitative number and can be
compared in terms of emotion-color distance. Based on experimental results, we
conclude that the proposed method gives results that are consistent with users’
perception at a high level. The proposed method is superior to traditional
methods of image annotation with words, in that, instead of just labeling with
words, the similarity of keywords and color image contents can be expressed in
terms of a quantitative number.
ปีที่พิมพ์
แหล่งทรัพยากรต้นฉบับ
ผู้ถือสิทธิ์
เงื่อนไขการใช้ข้อมูล
Copyright