At present, rapid, nondestructive, and objective identification of unqualified salted sea cucumbers with excessive salt content is extremely difficult. Artificial identification is the most common method, which is based on observing sea cucumber deformation during recovery after applying-removing pressure contact. This study is aimed at simulating the artificial identification method and establishing an identification model to distinguish whether the salted sea cucumber exceeds the standard by means of machine vision and machine learning technology. The system for identification of salted sea cucumbers was established, which was used for delivering the standard and uniform pressure forces and collecting the deformation images of salted sea cucumbers during the recovery after pressure removal. Image texture features of contour variation were extracted based on histograms (HIS) and gray level cooccurrence matrix (GLCM), which were used to establish the identification model by combining general regression neural networks (GRNN) and support vector machine (SVM), respectively. Contour variation features of salted sea cucumbers were extracted using a specific algorithm to improve the accuracy and stability of the model. Then, the dimensionality reduction and fusion of the feature images were achieved. According to the results of the models, the SVM identification model integrated with GLCM (GLCM-SVM) was found to be optimal, with accuracy, sensitivity, and specificity of 100%, 100%, and 100%, respectively. In particular, the sensitivity reached 100%, demonstrating an excellent identification ability to excessively salted sea cucumbers of the optimized model. This study illustrated the potential for identification of salted sea cucumbers based on pressure contact by combining image texture of contour varying with machine learning.

中文翻译：

---

不合格盐渍海参鉴定的新方法：在压力接触下图像纹理和机器学习的集成

目前，快速，无损，客观地鉴定含盐量过多的不合格盐渍海参非常困难。人工识别是最常用的方法，该方法基于观察施加-去除压力接触后恢复过程中海参的变形。本研究旨在模拟人工识别方法并建立识别模型，以借助机器视觉和机器学习技术来识别盐渍海参是否超出标准。建立了盐渍海参的鉴定系统，该系统用于传递标准的压力和均匀的压力，并收集盐渍海参在去除压力后的恢复过程中的变形图像。基于直方图（HIS）和灰度共生矩阵（GLCM）提取轮廓变化的图像纹理特征，分别结合通用回归神经网络（GRNN）和支持向量机（SVM）建立识别模型。使用特定算法提取盐渍海参的轮廓变化特征，以提高模型的准确性和稳定性。然后，实现了特征图像的降维和融合。根据模型的结果，发现与GLCM集成的SVM识别模型（GLCM-SVM）是最佳的，其准确性，灵敏度和特异性分别为100％，100％和100％。特别是灵敏度达到100％对优化模型过咸的海参表现出出色的识别能力。这项研究通过将轮廓变化的图像纹理与机器学习相结合，说明了基于压力接触识别盐渍海参的潜力。