【摘要】 [目的]提高自动化生产线中番茄品质检测的准确性与效率，解决传统检测依赖人工、精度低、效率差的问题。[方法]基于番茄自动化生产线品质检测系统，构建高光谱与机器视觉融合的内外品质检测系统。高光谱检测数据经预处理后，输入改进最小二乘支持向量机LSSVM模型检测番茄可溶性固形物和硬度，完成内部品质检测。机器视觉采集数据预处理后，输入改进YOLOv12模型检测外部缺陷，并计算番茄尺寸与果形指数，实现外部品质检测。并通过试验验证方法的优越性。[结果]内部品质检测方法对可溶性固形物、硬度预测决定系数（R~2）分别为0.965和0.975，均方根误差（RMSE）分别为0.082°Bx和0.061 N。改进YOLOv12模型缺陷检测平均精度均值为99.20%，检测速度>100帧/s，综合性能优于单一检测和现有方法。[结论]该融合检测系统可实现番茄内外品质同步、无损、高效检测，满足生产线需求。更多还原

【Abstract】 [Objective] To improve the accuracy and efficiency of tomato quality inspection in automated production lines, and solve the problems of traditional inspection, such as reliance on manual labor, low precision, and poor efficiency. [Methods] Based on the quality inspection system of the tomato automated production line, an integrated internal and external quality inspection system is developed by combining hyperspectral imaging and machine vision technologies. After preprocessing the hyperspectral detection data, an improved least squares support vector machine(LSSVM) model is employed to detect the soluble solids content and hardness of tomatoes, thereby completing internal quality inspection. For external quality inspection, the collected machine vision data are preprocessed, and an improved YOLOv12 model is utilized to detect external defects. Additionally, the size and fruit shape index of tomatoes are calculated. The superiority of the method is validated through experimental testing. [Results] The internal quality inspection method demonstrates high predictive accuracy, with determination coefficients(R~2) of 0.965 for total soluble solids and 0.975 for hardness, and root mean square errors(RMSE) of 0.082 ° Bx and 0.061 N, respectively. The improved YOLOv12 model achieves an average defect detection accuracy of 99.20% and a detection speed exceeding 100 frames/s. The overall performance of this integrated system is superior to that of single-detection approaches and existing methods. [Conclusion] This integrated detection system enables synchronous, non-destructive, and efficient detection of both internal and external quality of tomatoes, effectively meeting the requirements of automated production lines.更多还原