基于机器视觉的马铃薯黑心病检测机构设计与试验

Design and experiment of detection mechanism for potato blackheart based on machine vision

马铃薯黑心病是一种难以用肉眼分辨的内部病害。在生产加工中,为了准确检出黑心病个体,该研究基于光在马铃薯块茎组织中的透射特性设计了一种马铃薯黑心病检测机构。机构主要包括马铃薯块茎承托槽、光源单元以及光源位置调节轨道等部分,光源单元包括2个LED面光源,形成多点透射的结构,光源波长为705 nm。文中取50个正常样品和29个黑心样品采集图像分析,经过预处理,分别提取马铃薯目标图像和马铃薯高灰度区域的灰度平均值,以二者的相对比值作为黑心病的判别参数。通过与单点透射模式所采集的正常样品图像的相对比值比较,由该机构所采集的正常图像的相对比值普遍偏小,表明所获得图像的灰度均匀性更好,同时得到光源较佳夹角为45°。在该机构下获得的正常样品和黑心样品的灰度平均值相对比值的分类阈值为0.220,得到2种样品的判断正确率分别为98%和96.6%,且黑心样品的相对比值更大,表明使用该机构可以检测黑心马铃薯样品。该研究为马铃薯黑心病检测机构的设计提供了参考。

Potato blackheart is a kind of disease that happens in the inner of potato tuber.The potato with blackheart is difficult to be distinguished from the surface with the naked eye,which affects the quality of potato processed products seriously.In order to detect the potato with blackheart on the production line,a detection mechanism for potato blackheart was designed based on the light transmission properties of potato tuber internal organization in this study.The detection mechanism was composed of the supporting groove,the light source unit,the cooling fin,and the base.Firstly,the characteristics of size and morphology of potato tuber were analyzed based on the statistics rule.The potato tuber studied in this paper was round or elliptic fitted from the side direction,and the fitting diameter range was 40.0-100.0 mm.Then the light source wavelength was chosen as 705 nm according to the result of the research about potato black heart’s high spectral transmission detection.Two area light sources were used in the detection mechanism,constituting multipoint transmission structure.The mechanism was designed against the fitted circle of potato tuber sample in the first place.Two common tangent lines for any size of potato tubers were defined,and the locations of 2 tangent points on each fitting circle were set as the light source locations.The value of the included angle formed by the radius passing the tangent point and the vertical centerline was a determinant to the light source location.A proper included angle could lead to uniform illumination and less light leaking to the potato sample,and improve the consistency of imaging in the camera.The angle value was chosen as 45°by the analysis in this study.After that,the structure of the supporting groove was obtained.The height range between the center of the fitted circle of potato tuber and the bottom of the supporting groove was 20.0-62.43 mm.It could be accepted as a certain range to ensure the consistency of imaging.At the same time,the energy of the area light source installed on the supporting groove was completely absorbed by the potato tuber for its horizontal projection was exactly in the range of potato tuber horizontal radius.When the mechanism was used in the detection of fitted ellipse of potato tuber,the light source position,the included angle and the size of the supporting groove all met the detection demands by the analysis.In order to test the performance of the detection mechanism,a batch of fresh potato samples were chosen and sealed in zip-lock bags,then stored in an incubator at the temperature of 37°.Twenty-nine black heart samples were obtained after 24 d.79 potato tuber samples,including 50 normal samples and 29 black heart samples were chosen to collect the transmission images.The transmission images of 50 normal potato samples under single-point transmission mode were collected at the same time.Then the samples were sealed in zip-lock bags and stored in an incubator at the temperature of 37°.They were taken out and the light transmission images were collected by the detection mechanism after 24 d.Twenty-nine blackheart samples were obtained through the verification.After the background elimination processing with MATLAB,the potato tuber target image and the high grayscale value area after binarization and masking were extracted.The value of their average grayscale value relative ratio was acquired to distinguish the uniformity of the light transmission and potato black heart.The discriminant accuracy of the normal and black heart samples were 98%and96.6%,respectively.Results showed that,the mechanism introduced in the paper could be used to detect potato black heart.A reference is provided by the design of the detection mechanism in this study.