苹果果梗和缺陷的识别技术研究

Identification technique for stem-end and blemish in apples

为解决苹果自动检测和分选中苹果果梗和缺陷的识别这一瓶颈问题,提出基于HIS颜色模型下亮度I的分布特性,从信号处理的角度确定苹果的果梗和缺陷区域 再提取区域纹理特征,构筑模拟退火神经网络作为模式分类器,区分果梗和缺陷 试验结果表明用该方法识别准确率接近90%

A new method was developed to identify stem-end and blemish in apples precisely. Stem-end and blemish in apples were determined using one-dimension signal processing method based on HIS color model. By taking texture parameters of those two areas as input, a simulated annealing neural network was developed to identify stem-end and blemish in apples. Results show that the new technique is precise and effective in identification, and is feasible for practical apple sorting.