基于高光谱图像检测禽蛋新鲜度的研究

摘要

对禽蛋的新鲜度、贮藏期进行无损检测,在禽蛋生产、流通、加工领域具有重要意义.为了检测禽蛋的新鲜度,该文构建了高光谱图像采集系统,在400~1000 nm范围内获取245枚禽蛋贮藏42天的高光谱图像.研究利用连续投影算法提取特征波长检测禽蛋新鲜度,并由此建立基于特征波长的禽蛋新鲜度检测模型.研究结果表明：基于特征波长的禽蛋新鲜度模型的决定系数和剩余预测偏差分别达到0．87和2．80,说明高光谱图像技术可以检测禽蛋新鲜度.%In order to detect egg freshness incubation, a laboratory hyperspectral imaging system was set up to cap-ture hyperspectral transmission images of 245 eggs preserved for the forty-two days in the spectral region of 400~1 000 nm. The successive projections algorithm ( SPA ) was applied for effective wavelengths selection, and the freshness detection models based on the effective wavelengths were made with the prediction determination coeffi-cient of 0 . 87 , and RPD of 2 . 80 . This research demonstrates that the hyperspectral imaging technique is feasible for detecting eggs freshness.