塔里木大学学报

为快速获取红枣体积参数，提升红枣体积分级精度，本研究提出了一种边缘轮廓定积分测量红枣体积的方法。该方法通过对采集到的红枣图像进行灰度化、分割填充的预处理，将处理好的图像沿主轴旋转对齐分割为左、右两部分，采用八邻域追踪法检测提取两部分边缘轮廓有效像素点，并利用特定转换系数转换为实际轮廓坐标点，使用多项式对实际轮廓坐标点进行定积分，获得两部分体积，对齐合并求取完整体积。本研究分别对350颗红枣进行体积测量，将测量值与实测值进行线性拟合与误差分析，分析结果R=0.945 85,RMSE=3.134 mL,平均误差为11.57%。利用本方法对红枣进行体积分级，其分级正确率均在75.71%以上，平均分级正确率为80.57%。证明基于边缘轮廓定积分测量红枣体积具有一定的可行性，为红枣体积分级提供理论依据。

In order to obtain the volume parameters of jujube quickly and improve the precision of jujube volume classification, a method of measuring jujube volume by definite integral of edge contour was proposed. In this method, the collected jujube image was pre-processed by gray-scale, segmentation and filling, and the processed image were rotated and aligned along the main axis and divided into left and right parts. The effective pixel points of the edge contour of the two parts were detected and extracted by eight-neighborhood tracking method, and converted into actual contour coordinate points by specific conversion coefficient. The actual contour coordinate points were fixed integral by polynomial. The volume of two parts is obtained, and the full volume is obtained by alignment and merging. In this study, the volume of 350 red dates was measured one by one. Then, linear fitting and error analysis were conducted using the measured values and the true values. The analysis result showed R=0.945 85, RMSE=3.134 mL, and the average error was 11.57%. By using this method, the accuracy of jujube volume classification was over 75.71% and the average accuracy was 80.57%. It is proved that it is feasible to measure the volume of jujube based on the definite integral of edge contour, and provides a theoretical basis for the volume classification of jujube.