¹Graduate School of Information Science and Technology, Graduate School of Osaka Institute of Technology, Osaka Japan.
²Faculty of Information Science and Engineering, Ritsumeikan University, Shiga, Japan.
³Faculty of Information Science and Technology, Osaka Institute of Technology, Osaka Japan.
….

---

Abstract—The use of social network services (SNS) have become widespread in recent years, and it is used to express the situation of individual users by using short sentences and images. Especially, young people prefer to express themselves by using short sentences on Twitter and pictures on Instagram. However, CAMPUS LIFE DATA 2017 suggested that young people do few read a long sentence in a newspaper, journal, and more. We think that it is important for people who have not read long sentences to express long sentences with images. Therefore, we aim to express images and short sentences in SNSs from a long sentence in a newspaper, journal, and other reading material in this research. Specifically, we extract the impression of a person looking at an image or after reading a sentence. We calculate the similarity of each extracted word and generate a new image similar to the impression of the sentence. The results indicate that the system could estimate the impression of human beings. However, it was found that a large number of impression words and images are required to improve accuracy.

Index Terms—Impression words, image processing, morphological analysis, text to image.

[PDF]

Cite: Shohei Namisato, Kazuaki Yokota, Hiroki Yamaoka, Sho Ooi, Mutsuo Sano, "Generation of Image by Sentence Based on Impression Words in Image," Journal of Computers vol. 14, no. 5, pp. 349-358, 2019.