自动点卤槟榔图像识别方法研究

Research on recognition method for automatic orientating betel nut

针对机器对槟榔点卤出现漏点或者点空的情况,提出了一种基于H分量的图像分割方法来识别槟榔。将采集到的RGB彩色图像进行伽马增强再转换至HSV颜色空间,分离HSV颜色空间得H通道图像,再对H分量图像进行Otsu图像分割处理,结合图像形态学和区域生长法去除二值图像中出现的孔洞和小连通区域,最后通过绘制矩形框操作对识别出的槟榔进行标注。结果表明,该方法能完整地将槟榔从背景中分割出,且不会出现过度分割或者分割不足的问题。

When the betel nut bittern is in the process of betel nut bittern,the betel nut needs to be loaded first.In this process,there will be some places in the wobble plate that are not equipped with betel nut.In turn,it will cause the machine to perform bittern on empty positions,causing waste of bittern and reducing the efficiency of bittern.In view of the situation that the machine is leaking or emptying the betel nut on the betel nut swing plate,it is necessary to identify the betel nut on the bittern plate before the bite point.Due to the large difference in color between the betel nut and the betel nut wobble plate,in order to be able to completely segment the betel nut from the wobble plate without over-segmentation or under-segmentation,an image segmentation method based on H component is proposed to identify betel nut.In this method,the acquired RGB color image is first gamma-enhanced and then transferred to the HSV color space,the HSV color space is separated,and the H channel image is obtained.The H component image is segmented by the Otsu image segmentation method,combining image morphology and regional growth The method removes the holes and small connected areas in the binary image,and finally marks the identified betel nuts by drawing a rectangular frame.Experimental results show that using this method can completely separate betel nuts from the background,without over-segmentation or under-segmentation,Thus accurately identifying betel nuts.