基于质构检测的仿指压触生物力与海参体形协同响应机制研究

Texture is one of the most important quality indexes of sea cucumber. Detection nature based on tactile perception is the complex association of tramolecular and intermolecular stress and strain field. Interaction between deformation and pressing force was avoided through unified specification. Currently, the response mechanism between the two is not clear, and the deformation interference for texture detection can’t be eliminated. The sea cucumber with complex profile is applied to the research. At the micro level, the relationship among collagenous fiber ,pressure and deformation is revealed. Bionic cooperative response detection system is built through sensory evaluation simulating. Bionic finger pressing force and deformation are represented by biomechanical force and serial images which are exported from the system. Specific to the profile corresponding to instantaneous biomechanical force, the scale invariant relative morphological features are extracted. The fusion model of time and spatial domain texture information is established. Collaborative response relationship of biomechanical force and deformation are illuminated. Specific to the contraction and resilience process of sea cucumber, through graphics energy and two-dimension discrete wavelet texture analysis, kinematics regularity of figure is excavated, and the law of mapping between figure response and texture parameters is revealed. On this basis, the information of pressing biomechanical force of bionic finger and dynamic figure are fused to build fuzzy detection model. Based on texture detection, the collaborative response mechanism in multiple domain is established. The new idea for food texture detection of complex figure is proposed.

质构是食品最重要的品质指标之一。基于压触的检测本质为分子内和分子间应力、应变场的复杂关联关系，目前通过统一规格规避形变与触力的相互影响，致使二者的协同响应机制尚不清晰，各向异性形变响应对检测的干扰无法消除。本项目以形状复杂的海参为研究对象，在微观层面明确胶原纤维与受力、形变的关联性，以感官评定为仿生模板构建检测系统，输出生物力和时间序列图像以表征仿指压触和体形的协同响应。面向与生物力具有时间对应性的体形轮廓，分析并提取尺度不变的相对形态学特征，建立时、空域融合的质构信息模型，明确生物力与体形协同响应关系；针对海参压缩和回弹的动态变化，利用图形能量和二维离散小波纹理分析，挖掘海参体形响应运动学规则，揭示体形响应与质构特性的映射规律。在此基础上，通过模糊网络构建生物力与体形信息的质构检测模型，建立仿指压触生物力与体形协同响应检测机制，为具有复杂边界形状的食品质构无损检测提供新思路。

质构是食品最重要的品质指标之一。基于压触的检测本质为分子内和分子间应力、应变场的复杂关联关系，目前通过统一规格规避形变与触力的相互影响，致使二者的协同响应机制尚不清晰，各向异性形变响应对检测的干扰无法消除。本项目面向盐渍海参质构检测，以感官评定为仿生模板，构建了基于协同响应的仿生检测系统，实现了仿指压触力作用和海参样品回复过程的图像自动采集与处理。面向动态能量特征图，提出了基于灰度共生矩阵的体形图像特征提取方算法；同时建立力学统计特征、做功特征的提取算法；进行体形图像特征、力学特征与质构特性相关性分析，确定了用于生物力与海参体形协同响应检测的显著特征。分别面向单一的体形图像和力学特征，利用BP和广义回归（GRNN）神经网络，构建海参质构特性预测模型，挖掘海参体形响应、力学响应与质构特性的映射规律。在此基础上，利用串行和主成分分析法，提出了生物力和体形特征降维与融合方法，建立了生物力与海参体形协同响应质构预测模型，揭示了基于质构的生物力与海参体形协同响应检测机制。针对硬度、弹性和咀嚼性的质构特性预测，基于GRNN的模型效果最佳，模型预测精度分别可达0.90、0.92和0.86。