To search, Click below search items.


All Published Papers Search Service


Efficient Retrieval on Dense Vector by Similarity Preserve Hash in Vegetable Geographical Origin Identification System


Nobuyoshi Sato, Minoru Uehara, Koichiro Shimomura, Hirobumi Yamamoto, Kenichi Kamijo


Vol. 6  No. 10  pp. 227-235


Recently, camouflaging geographical origin of agricultural products is a major problem in Japan. Therefore, we developed a distributed geographical origin identification system which identifies cultivated farms of vegetables by using trace element compositions of vegetables. This system stores compositions of trace, or very small quantities of elements into database which located on farming districts, and compares them to trace element compositions of vegetables which gathered from food distribution channel such as markets, food factories. The comparison is done by calculating correlation coefficient. This system can be considered as an information retrieval system which gives only an answer of which trace element composition is similar to given query vegetable’s trace element composition. In this system, trace element compositions are expressed as dense vector, and they are retrieved by one-to-one comparison. In this paper, we describe applying Similarity Preserve Hash (SPH) into our geographical origin identification system, with assumption on data disposition.


Similarity Preserve Hash (SPH), Retrieval on dense vector