Note
Image processing for evaluation of papilla clearness of dried sea cucumber products
2021 年 27 巻 1 号 p. 69-74
詳細
2021 年 27 巻 1 号 p. 69-74
We aimed to establish an objective method to evaluate the quality of dried sea cucumber products. Clearness of papillae on the body wall affects the quality and commercial value of this luxury seafood. Unfortunately, an objective method for evaluating papilla clearness remains to be developed. We digitalized clearness of papillae from dried sea cucumber products images using TouchDeMeasure. This software application was developed by the authors. Using this tool, we measured body length (BL), total perimeter including all papillae (TP), and ellipse perimeter excluding the papillae (EP). Results showed that TP/BL, TP/EP, and (TP-EP)/BL values for a clear papilla were significantly higher than those for an unclear papilla. Additionally, we confirmed that these parameters corresponded with the grading levels used by a connoisseur. These experimental results show that our method digitally analyzes the clearness of papillae to evaluate the quality of dried sea cucumber products.
The sea cucumber, Apostichopus japonicus, is an important seafood in Japan. Dried sea cucumber product is a traditional processed food that is prepared by boiling and drying these marine organisms. More than 80% of the sea cucumbers harvested globally are processed into dried products (Xu et al., 2010). The dried products are mainly exported to China as an expensive luxury food. Hokkaido dried sea cucumber products, in particular, have been traded at a high price (Akamine, 2009).
Truong and Le (2019) measured the length of papillae of six types of dried sea cucumber products using a digital caliper. They reported that the papillae of Hokkaido products were longer than those of Mexico, Kansai, and Aomori products. They proposed that long papillae might be one of the factors contributing to the high commercial value of Hokkaido products. However, dried sea cucumber products have a complex external form. The papilla length is not the lone factor contributing to perceived quality. Generally, the perceived quality of dried sea cucumber products relies on characteristics of the papillae, including their length, shape, sharpness, and distribution, as these features determine visual appeal. Papilla visibility, termed papilla clearness (or “Ibodachi”), is linked to the market value of dried sea cucumbers products — the clearer the papillae, the higher the commercial value of the product. Despite the link between clearness of papillae on the body wall and product value, a scientific method to evaluate papilla clearness has not been developed. Consequently, quality evaluation of dried sea cucumber products is currently done subjectively by individual evaluators such as producers, wholesalers, retailers and other buyers.
Previous studies on the quality of dried sea cucumber products have focused mainly on chemical components and microstructure (Fukunaga et al., 2002), soaking and texture (Fukunaga et al., 2004), inorganic elements and volatile organic compounds (Lee et al., 2013), and comparative chemical characterization (Thu et al., 2019). Few studies have focused on evaluating quality based on the features of the papillae. Based on these circumstances, we digitalized the clearness of papillae of dried sea cucumber products using image processing. Our aim is to establish an objective method for evaluating the quality of dried sea cucumber products.
Dried sea cucumber samples Our investigation used dried sea cucumbers products processed in a seafood factory in Yoichi. These products were processed in June 2018. From these products, we selected 13 typical dried sea cucumber products with clear papillae (superior products) and 15 typical sea cucumber products with unclear papillae (inferior products), as shown in Fig. 1. The dried sea cucumber products were photographed using a digital camera (E-PL2, OLYMPUS). The digital camera, along with a fixed photography platform and a white board (as a background), was placed in a seafood factory. To compare the influence of photographic angle on the image processing results, photographs of the dried sea cucumber products were taken, focusing on the overhead and lateral angles (Fig. 1).
Images of dried sea cucumber product: (A) Image of a superior product taken from overhead angle; (B) Image of an inferior product taken from overhead angle; (C) Image of a superior product taken from lateral angle; (D) Image of an inferior product taken from lateral angle.
Selection of parameters and image processing software Using the images of the dried sea cucumber products, we measured body length (BL), total perimeter (TP), and ellipse perimeter (EP). BL was measured as the length from the anterior to the posterior end, TP was measured as the perimeter including all papillae, and EP was measured as the perimeter excluding the papillae. In our study, we noted three parameters: TP/BL, TP/EP, and (TP-EP)/BL. Data were measured using the TouchDeMeasure (Ver.0.7.1.0) software.
TouchDeMeasure was developed by the authors to measure the size of marine products, such as sea cucumbers, clams, and urchins (Enomoto and Toda, 2018; Nankumo et al., 2016). TouchDeMeasure extracts the object area in the image based on the user's specifications and automatically measures the area, color, size, and other features. In the case of dried sea cucumber products, the object region is extracted and the shape information is quantified by the Fourier descriptor (which is expressed in a frequency domain).
Information on the dried sea cucumber products region was extracted using the region expansion method (Enomoto and Toda, 2018). This method divides the region into seed and neighboring color information to maximize the evaluation value. A brief description of the method to acquire information on the dried sea cucumber products region without the papillae is provided. The contour of the dried sea cucumber products is defined by a closed curve of N points. At every point t, there is a complex z(t), 0 ≤ t < N-1 (Granlund, 1972). When j is an imaginary unit, the discrete Fourier expansion equation used is:
 |
 |
Therefore, we defined the shape of the dried sea cucumber products papillae as a high-frequency component, and removed it once we obtained the shape information. In the shape data, the remaining low-frequency component corresponds to EP. Information on the dried sea cucumber products region without papillae is obtained by removing the Fourier coefficients that are higher than the threshold Thk and by restoring them by an inverse Fourier transform. In this study, we set the threshold for Thk at 7. This information is used to measure the dried sea cucumber products region without papillae. The body length (BL) is the length of the skeleton (centerline) of the obtained region (Nankumo et al., 2016).
Comparison of our parameters with grading by a connoisseur Consistency of the three parameters in relation to sea cucumber quality, when compared with grading by a connoisseur, is important for the practical application of our method. Therefore, for this part of the study, we compared our selected parameters with the grading of dried sea cucumber products by a connoisseur. We used dried sea cucumber products produced in June 2018 at a seafood factory (Hiyama Fisheries Cooperative Association). These products were graded into three quality levels by the connoisseur. He was a plant manager at the factory and had sufficient ability and experience to grade the products. Based on a subjective evaluation of clearness of papillae, the dried sea cucumber products were graded as superior product (n = 18), normal product (n = 13), and inferior product (n = 13) (Fig. 2). We obtained BL, TP, and EP from images of the dried sea cucumber products in each grade using TouchDeMeasure and calculated the three parameters.
Images of dried sea cucumbers product graded into 3 levels by a connoisseur: (A) superior product; (B) normal product; (C) inferior product. All images obtained from overhead angle.
Statistical analysis Statistically significant differences between the parameters obtained in this experiment were investigated. The statistical difference between two groups was determined by Welch's test. The statistical difference between three groups was determined by one-way analysis of variance (ANOVA). If the one-way ANOVA was significant, differences between individual groups were estimated using Scheffe's multiple comparison procedure. Differences with p < 0.01 were considered significant. In two groups, to clarify the difference between the two groups, effect size (d) was calculated using the following equation (McGrath and Meyer, 2006):
 |
where X̅a and X̅b denote the means for two sample groups and Sa and Sb denote the standard deviations for two sample groups.
Selection of parameters We compared combinations of the data obtained using TouchDeMeasure. Our results showed that for all three parameters: TP/BL, TP/EP, and (TP-EP)/BL, the values for the superior samples were significantly higher than those for the inferior samples (p < 0.01) (Fig. 3). These results suggest that these three parameters are positively correlated to the degree of papilla clearness. There are possible explanations for the superior samples that resulted in higher values compared to the values observed with inferior samples. The TP of the superior samples was longer than that of the inferior samples because the dried sea cucumber products in the clear papillae samples had longer and more numerous papillae. Additionally, a similar result was obtained for TP-EP, which represented the perimeter length of the papillae only. The value for TP-EP of the superior samples was greater than that for the inferior samples. This is a logical result. However, TP and EP are also influenced by the product size. To compare the differences between the two groups, a standard length is required for each product. In this regard, both the BL and EP were considered adequate to be used as a standard.
Box plot diagram of parameters obtained from dried sea cucumber product images.
(A) TP/BL, overhead angle; (B) TP/BL, lateral angle; (C) TP/EP, overhead angle; (D) TP/EP, lateral angle; (E) (TP-EP) /BL, overhead angle;(F) (TP-EP) /BL, lateral angle; Different letters indicate significant (p < 0.01, Welch test, n = 28).
Angle of photography To confirm the best angle of photography for image analysis, we calculated the effect size (as shown in Eq. 3) for both the superior and inferior samples for each photographic angle. We found that the size effect of the overhead angle was clearly greater than that of the lateral angle (Table 1). The result was the same for TP/BL, TP/EP, and (TP-EP)/BL. Effect size is the magnitude of the difference between the intervening groups (Sullivan et al., 2012). This means that the larger the effect size, the greater the magnitude of the difference between the two groups. These results revealed that images of sea cucumber products photographed from an overhead angle showed the differences more clearly between the two groups than the images taken from a lateral angle. Yaqing et al. (2012), reported that the number of dorsal papillae was markedly higher than that of the ventral papillae. Consequently, images of the dried sea cucumber products taken from a lateral angle did not show as many of the dorsally located papillae (in silhouette) as compared to the images taken from an overhead angle.
| TP/BL | TP/EP | (TP-EP) / BL | |
|---|---|---|---|
| Overhead angle | 3.22 | 3.31 | 3.30 |
| Lateral angle | 1.77 | 1.86 | 1.86 |
Comparison of our parameters with grading by a connoisseur As shown in Fig. 4, the parameter means significantly corresponded with the grading order: superior > normal > inferior (one-way ANOVA, Scheffe's multiple comparison procedure, p < 0.01) and the plotted data for the superior and inferior product levels were consistently separated. Similar results were obtained for all three parameters (TP/BL, TP/EP, and (TP-EP)/BL). These results revealed that all three parameters corresponded with the subjective grading by a connoisseur. However, one outlier was observed in the normal grade samples. Since this outlier occurred from the same product, this sample might essentially be selected as superior grade. If this outlier was excluded, the plotted data separated completely between the superior and normal grades. In this regard, our method was considered an effective tool, especially for selecting superior grade products. In practice, judging the difference between superior and inferior grade products is not difficult. However, the difference between superior and normal grade products is not as evident as the differences between products in the superior and inferior grades.
Box plot diagram of parameters obtained from dried sea cucumber product images grading into 3 levels by a connoisseur. (A) TP/BL; (B) TP/EP; (C) (TP-EP) / BL, Different letters indicate significant (p < 0.01, Scheffe multiple range test, n = 44.)
To determine optimum parameters, we calculated the effect size (as shown in Eq. 3) between superior, normal, and inferior products for each parameter. The results showed that the effect size between superior and normal products of TP/EP and (TP-EP)/BL were slightly higher than those of TP/BL. In contrast, the effect size between normal and inferior grade products of TP/BL was slightly higher than that of TP/EP and (TP-EP)/BL.
The observations between inferior and superior grade products were the same as the results of normal and inferior grade products (Table 2). These results suggest that we must choose an optimum parameter according to the circumstances. In particular, the parameter that includes BL needs further examination. It is well-known that the body length of a sea cucumber is variable due to its extendible and contractible body shape (Yamana and Hamano, 2006). We have empirically observed that a sea cucumber that has lost its freshness will have a tendency to slack (causing the BL to increase). Therefore, it is possible that a parameter that includes BL will assist us in understanding the relationship between freshness of fresh sea cucumber and clearness of papillae of dried sea cucumber products as we endeavor to identify an objective technique to evaluate dried product quality.
| Superior and normal | Normal and inferior | Inferior and superior | |
|---|---|---|---|
| TP / BL | 3.27 | 2.24 | 5.15 |
| TP / EP | 3.31 | 2.20 | 5.11 |
| (TP-EP) / BL | 3.31 | 2.21 | 5.10 |
Acknowledgements We would like to thank to Mr. Tsunehiko Kaneda, Kaneda fishery Co. Inc. and Mr. Takasumi Hinuma, Hiyama Fisheries Cooperative Association, for kindly providing samples. We also thank Ms. Yukie Kawamura for assistance with the image analysis.
