公开(公告)号：WO2017194398A1

公开(公告)日：2017-11-16

申请号：PCT/EP2017/060750

申请日：2017-05-05

Applicant: BAYER CROPSCIENCE AKTIENGESELLSCHAFT

Inventor： KIEPE, Björn ,  SCHILLING, Thomas ,  FOIS, Franco ,  RECHSTEINER, Daniel ,  HELLWEG, Sebastian ,  GLADBACH, Alexandra ,  STEPPONAT, Birgit

IPC: G06K9/00 ,  G06K9/03 ,  G06K9/46 ,  G06K9/62 ,  G06T7/00 ,  G06T7/40 ,  A01M21/04 ,  G06T7/10

CPC classification number: G06K9/00671 ,  A01M21/043 ,  A01M99/00 ,  G06K9/036 ,  G06K9/4642 ,  G06K9/4652 ,  G06K9/627 ,  G06K2209/27 ,  G06T11/001

Abstract: A method (100) for recognizing weed in a natural environment may be provided. A digital image of weed in an early development stage is captured (102) among cultivated crop. Areas with a predefined color and texture specification are contoured (104) defining a boundary contour. The weed is displayed (106) together with a frame and a quality indicator. Moreover, the digital image is stored (110) only if a predefined quality criterion is met. Then, unnecessary color information may be reset (112), the digital image is sent (116) for a further examination, and a weed name of the weed of the captured image and a related probability value indicative of a probability of a match between the weed name and the weed of the captured digital image is received (118).

Abstract translation: 可以提供用于识别自然环境中的杂草的方法（100）。 在早期发育阶段的杂草数字图像被捕获（102）在栽培作物中。 具有预定颜色和纹理规格的区域被轮廓化（104）以定义边界轮廓。 显示杂草（106）以及框架和质量指标。 此外，只有在满足预定质量标准的情况下，数字图像才被存储（110）。 然后，不必要的颜色信息可以被重置（112），数字图像被发送（116）以用于进一步的检查，并且捕获图像的杂草的杂草名称以及表示捕获图像之间的匹配概率的相关概率值 接收到杂草名称和捕获的数字图像杂草（118）。

公开(公告)号：WO2017194399A1

公开(公告)日：2017-11-16

申请号：PCT/EP2017/060751

申请日：2017-05-05

Applicant: BAYER CROPSCIENCE AKTIENGESELLSCHAFT

Inventor： KIEPE, Björn ,  SCHILLING, Thomas ,  GLADBACH, Alexandra ,  STEPPONAT, Birgit ,  FOIS, Franco ,  RECHSTEINER, Daniel ,  HELLWEG, Sebastian

IPC: G06K9/00 ,  G06K9/46 ,  G06K9/62 ,  G06T7/11

CPC classification number: G06K9/00671 ,  G06K9/4638 ,  G06K9/4652 ,  G06K9/627

Abstract: A method for identifying of a type of weed in a natural environment may be provided. The method comprises receiving a digital image of the weed in an early growth stage together with related metadata, contouring areas of leaves and determining whether the weed is a monocotyledon or a dicotyledon. Furthermore, the method comprises determining a growth stage and determining the type of weed identified by a weed name and a probability of the correctness of the determination using at least one out of a plurality of the received metadata and a plurality of metadata determined during the contouring, during the determining whether the weed is a monocotyledon or a dicotyledon, and during the determination of the growth stage of the weeds as input parameters to a set of classifiers using a storage comprising names of types of weeds together with a plurality of sets of metadata per weed type.

Abstract translation: 可以提供用于识别自然环境中的杂草类型的方法。 该方法包括在早期生长阶段与相关的元数据一起接收杂草的数字图像，叶的轮廓区域并确定杂草是单子叶植物还是双子叶植物。 此外，该方法包括确定生长阶段并且使用多个所接收的元数据中的至少一个和在轮廓处理期间确定的多个元数据来确定由杂草名称识别的杂草的类型以及确定的正确性的概率 在确定杂草是单子叶植物还是双子叶植物期间，以及在确定杂草的生长阶段期间，使用包括杂草类型名称和多组元数据的存储器作为输入参数给一组分类器 每种杂草类型。