X射线图像技术对红毛丹内部品质的检测

研究了利用X射线图像处理及模式识别方法检测红毛丹内部品质——可食率、可溶性固形物含量。首先用阈值分割法去除红毛丹背景，然后用模糊C均值聚类方法来分割果肉区域。红毛丹可食率以分割出的果肉区域像素个数与整个果实区域像素个数之比来表征，实验结果表明，误判率小于10％。红毛丹可溶性固形物含量的X射线图像检测采用纹理特征描述的方法，利用直方图矩特征和支持向量机回归方法预测红毛丹可溶性固形物含量的相关系数最高可达87．2％。

Detection of Internal Quality of Rambutan Fruit by Using X-ray Imaging

The methods of X-lay image processing and pattern recognition were used to test internal quality of rambutan fruit, the edible rate and the content of soluble solids. The rambutan flesh was segmented using the FCM (fuzzy C-mean) clustering method after removing the background of the image. Edible rate of the fruit was expressed by the ratio of the pixels of the segmented parts to the pixels of the whole fruit. The result showed that the error was less than 10 % in the experiment. Soluble solids content of rambutan was described by the image texture, and its maximum correlation coefficient was 87.2 % based on the support vector regression and histogram matrix characteristics.