본문 바로가기
  • Home

A Similarity-based Inference System for Identifying Insects in the Ubiquitous Environments

  • Journal of The Korea Society of Computer and Information
  • Abbr : JKSCI
  • 2011, 16(3), pp.175-188
  • Publisher : The Korean Society Of Computer And Information
  • Research Area : Engineering > Computer Science

전응섭 1 Chang, Yong Sik 2 권영대 3 김용남 4

1인덕대학
2한신대학교
3경기도산림환경연구소
4엠비즈테크 기술연구소

Accredited

ABSTRACT

Since insects play important roles in existence of plants and other animals in the natural environment, they are considered as necessary biological resources from the perspectives of those biodiversity conservation and national utilization strategy. For the conservation and utilization of insect species, an observational learning environment is needed for non-experts such as citizens and students to take interest in insects in the natural ecosystem. The insect identification is a main factor for the observational learning. A current time-consuming search method by insect classification is inefficient because it needs much time for the non-experts who lack insect knowledge to identify insect species. To solve this problem, we proposed an smart phone-based insect identification inference system that helps the non-experts identify insect species from observational characteristics in the natural environment. This system is based on the similarity between the observational information by an observer and the biological insect characteristics. For this system, we classified the observational characteristics of insects into 27 elements according to order, family, and species, and proposed similarity indexes to search similar insects. In addition, we developed an insect identification inference prototype system to show this study's viability and performed comparison experimentation between our system and a general insect classification search method. As the results, we showed that our system is more effective in identifying insect species and it can be more efficient in search time.

Citation status

* References for papers published after 2023 are currently being built.