基于最小外接矩形的牛角椒大小分级研究

为实现用机器取代人工对牛角椒进行大小分级，针对牛角椒的外形近似弧的特点，设计一个基于图像处理的牛角椒大小分选系统。研究选取100个某一品种的牛角椒，首先对牛角椒进行拍摄，然后将拍摄的图像进行图像预处理，提取辣椒边缘，并去除辣椒柄，最后对去柄后的边缘通过最小外接矩形的方法得到矩形的4个顶点坐标，再选取1个牛角椒中点，计算出牛角椒的弧形长度。将牛角椒的系统检测大小与实际大小进行比较，系统检测的辣椒分级正确率为95%，研究结果为牛角椒大小自动分级提供了参考。

Pepper Grading Based on the Minimum Bounding Rectangle

In order to realize the pepper size grading by machine, according to pepper shape that is approximate to an arc, the authors designed a sorting system of pepper based on image processing. 100 peppers were selected for the research, the peppers were first photographed, then the images were processed for extracting the edge of the pepper and removing the pepper’s handle, at the end, four vertex coordinates were calculated by the minimum bounding rectangle diameter on the edge of the peppers with removed handles. A mid-point of the pepper was selected to calculate the arc length of the pepper. A comparison of the measured results and the detection results showed that the separation accuracy was 95%. The study could provide certain reference for automatic pepper grading according to the pepper size.