ファジィ推論を用いたフグ肉の肉眼鑑別モデルに関する基礎的検討

Visual Discrimination Models of the Dressed Puffers（Migaki）by Skilled Cooks Using Fuzzy Inference

Creators Kogami Kenichiro

Creators Nakamura Makoto

Creators Ohta Hiromitsu

Creators Tokunaga Kazuhiro

Creators Maeda Toshimichi

Experienced puffer fish chefs can quickly distinguish between species and between natural and cultivated fish based on the appearance of dressed puffer fish. This study aims to construct a system of differentiating between puffer fish based on the visual cues used by experienced chefs. The species used were Tiger Puffer（Takifugu rubripes), Purple Puffer（T. porphyreus), Panther Puffer（T. pardalls), Vermiculated Puffer（T. snyderi）and White Chestnut Puffer（Laocephalus wheeleri). Moreover, both natural and cultured Tiger Puffer were used. Ten points on the bodies of the fish were selected to measure color, and color measurements were made up to 72 hours after the specimens were obtained. Next, the color difference（ΔE^*ab）between two species was calculated. Consequently, it was revealed that from 3 to 10 points are necessary to visually differentiate two species； this was consistent with an explanation provided by an experienced chef prior to the experiment. The results showed that chefs can visually differentiate between fish with extreme precision. We compared two species at a time, as well as all five species at once using discriminant analysis based on the color measurement points. The analysis showed the discrimination rate between two species to be ≥90.2% and the rate among all species to be 90.9%. The discrimination rates obtained had high probability. Finally, in the discrimination model using fuzzy inference, all models were in accordance with the results of inference and species discrimination. Therefore, models based on the visual cues used by chefs are useful in differentiating between puffer fish.

[ISSN]0370-9361