WO2018214054A1 - 一种视觉检测方法、设备、系统及具有存储功能的装置 - Google Patents
一种视觉检测方法、设备、系统及具有存储功能的装置 Download PDFInfo
- Publication number
- WO2018214054A1 WO2018214054A1 PCT/CN2017/085708 CN2017085708W WO2018214054A1 WO 2018214054 A1 WO2018214054 A1 WO 2018214054A1 CN 2017085708 W CN2017085708 W CN 2017085708W WO 2018214054 A1 WO2018214054 A1 WO 2018214054A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- feature value
- image feature
- abnormal
- correct
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
Definitions
- the present invention relates to the field of visual inspection technologies, and in particular, to a visual detection method, device, system, and device having a storage function.
- the inspection of the product is generally carried out by means of manual sampling.
- the labor intensity of manual inspection fatigue is prone to work for a long time, the quality of the inspection cannot be guaranteed, and the pass rate of the sampled product cannot be guaranteed.
- manual testing is difficult to adapt.
- the detection method based on the vision system can solve the above problems well and improve the detection efficiency.
- the work of the vision system in the industrial pipeline is sometimes prone to image anomalies due to camera errors, pipeline failures, or disturbances in the pipeline.
- there is no feedback mechanism for image abnormality When an image abnormality occurs, the entire pipeline operation enters an error state, and industrial application cannot be guaranteed.
- the technical problem to be solved by the present invention is to provide a visual detection method, device, system and device having the storage function, which can timely feedback the acquired image anomaly and prevent the industrial pipeline from entering an error state.
- a technical solution adopted by the present invention is to provide a visual detection method, including:
- a visual inspection device including a processor and a memory, the memory storing program data, and the processor loading the program data for performing the above method.
- another technical solution adopted by the present invention is to provide a device having a storage function, which stores program data, the program data being adapted to be loaded by a processor and to execute the above method.
- a visual inspection system including:
- a visual acquisition device for acquiring images in the current pipeline
- a visual inspection device configured to acquire an image in a current pipeline collected by the visual system, and determine whether an image feature value of the image is within a correct image feature value threshold;
- an output device configured to output an abnormality warning when the image feature value of the image is not within the correct image feature value threshold.
- the invention extracts the image feature value of the acquired image, determines whether the image feature value of the image is within the correct image feature value threshold value, outputs an abnormal warning or automatically adjusts the correct image feature value threshold value, can timely feedback the acquired image abnormality, and avoids the industrial pipeline. Enter the error state.
- FIG. 1 is a schematic flow chart of an embodiment of a visual inspection method according to the present invention.
- FIG. 2 is a schematic flow chart of another embodiment of a visual inspection method according to the present invention.
- FIG. 3 is a schematic diagram of an image currently acquired by another embodiment of the visual inspection method of the present invention.
- FIG. 4 is a schematic structural view of an embodiment of a visual inspection device of the present invention.
- FIG. 5 is a schematic structural diagram of an embodiment of an apparatus having a storage function according to the present invention.
- Figure 6 is a block diagram showing the structure of an embodiment of the visual inspection system of the present invention.
- an embodiment of the visual inspection method of the present invention includes:
- the vision system can be any vision system that can capture images in the current pipeline, such as a machine vision system.
- image acquisition is performed by an imaging device such as a camera or a mobile phone.
- an imaging device such as a camera or a mobile phone.
- the warning is output by means of sound or flashing, for example, by a buzzer, an alarm light, or by displaying an alarm on the screen.
- the embodiment of the invention extracts the image feature value of the acquired image, determines whether the image feature value of the image is within the correct image feature value threshold, outputs an abnormal warning or automatically adjusts the correct image feature value threshold, and can timely feedback the acquired image abnormality to avoid The industrial pipeline entered an error state.
- another embodiment of the visual inspection method of the present invention includes:
- the correct image samples are stored in the sample set to form a correct image sample set, and a correct image feature value threshold is generated according to the formed correct image sample set; the abnormal image sample is stored in the sample set to form an abnormal image sample set, and generated according to the formed abnormal image sample set
- the abnormal image feature value threshold, the abnormal image sample set includes an abnormal image alert class corresponding to different abnormal types.
- the correct image samples are pre-stored according to the correct criteria for the product on the assembly line
- the abnormal image samples are pre-stored according to the criteria of the product that may be abnormal.
- the image feature value is used to determine whether the image is abnormal and target detection of the image; wherein the image feature value may be average brightness, uniform average brightness, sharpness of the reference area around the detected target area, and the detected target area Gray scale difference or local gray scale distribution, etc.
- comparing the image feature values extracted in the image with the pre-stored correct image feature values determining whether the extracted image feature values are within a pre-stored correct image feature value threshold.
- the visual system determines that the image feature value of the image is not within the correct image feature value threshold, the image is an abnormal image, and the visual system outputs an abnormal warning.
- the image feature value of the image is added to the abnormal image sample set.
- comparing the image feature value extracted in the image with the pre-stored abnormal image feature value determining whether the extracted image feature value is within a pre-stored abnormal image feature value threshold; if the extracted image feature value is in the abnormal image Within the eigenvalue threshold, it is determined that the image is abnormal, and the image feature value of the extracted image is added to the abnormal image sample set of step S101 to perform incremental learning training, and the abnormal image feature value threshold is updated.
- the abnormal image sample set includes different types of abnormal image warning types, and different image early warning classes treat the images differently, for example, according to the collected abnormal images belonging to different early warning types, the visual system outputs an abnormal warning or automatic Adjust the correct image feature value threshold to avoid issuing an exception warning. If the visual system determines that the image feature value of the image is neither within the correct image feature value threshold nor within the abnormal image feature value threshold, it is determined as a new abnormal image, and the new abnormal image is stored and counted as a new abnormality. When the number of images reaches a threshold value, for example, the number reaches 100 times, a new abnormal image warning class is created from the 100 new abnormal images, and the image feature values of the images jointly extracted by the 100 new abnormal images are added to step S101. The abnormal image sample set is used for incremental learning training to update the abnormal image feature value threshold.
- the image feature value of the image is within the correct image feature value threshold, the image is subjected to target detection; the algorithm used for the target detection of the image is different from the algorithm used to determine whether the image is abnormal.
- the target detection if the target detection is successful, it is determined that the image is correct, and the detection result is directly output; if the target detection fails, it is determined that the image is abnormal, and the image feature value of the abnormal image is returned to the abnormal image as the learning input in step S601. Sample set.
- the extracted image feature value is within the correct image feature value threshold, and the target detection is successful, determining that the image is correct, and adding the image feature value of the extracted image to the correct image sample set of step S101 to perform Incremental learning training, updating the threshold value of the abnormal image feature value.
- the currently acquired image includes a target detection area 10 and a reference area 20.
- the correct image sample set and its correct image feature value threshold, abnormal image sample set and its abnormal image feature value threshold are pre-stored in the sample set.
- the image feature value threshold is updated to obtain the image feature of the current reference area.
- Value threshold for example, the gray value V1 of the reference area is obtained by learning, and the threshold value of the correct image feature value is 50 by learning, and the gray value of the reference area 20 in the currently acquired image is V2, when V2>V1+50, exceeds If the correct image feature value threshold is used, it is judged that the reference region 20 is abnormal in the feature value, and an abnormal image warning is output.
- the correct image feature value threshold can also be manually set, and no limitation is made here.
- the embodiment of the invention detects and analyzes the collected image, determines whether the image feature value of the image is within the correct image feature value threshold or the abnormal image feature value threshold, and outputs an abnormal warning, and uses the image feature value of the image as the learning input.
- the sample is added to the sample set, and the image feature value threshold is updated, and the acquired image anomaly can be fed back in time to avoid the industrial pipeline entering the error state, and the sample set is continuously updated to make the analysis result more accurate and improve the detection efficiency.
- an embodiment of the visual inspection device 40 of the present invention includes a processor 401 and a memory 402.
- the memory 402 stores program data, and the processor 401 loads program data for performing the visual inspection method described above.
- the embodiment of the invention extracts the image feature value of the acquired image, determines whether the image feature value of the image is within the correct image feature value threshold, outputs an abnormal warning or automatically adjusts the correct image feature value threshold, and can timely feedback the acquired image abnormality to avoid The industrial pipeline entered an error state.
- an embodiment of an apparatus 50 having a storage function of the present invention stores program data 501 adapted to be loaded by a processor and to perform the visual inspection method described above.
- the embodiment of the invention extracts the image feature value of the acquired image, determines whether the image feature value of the image is within the correct image feature value threshold, outputs an abnormal warning or automatically adjusts the correct image feature value threshold, and can timely feedback the acquired image abnormality to avoid The industrial pipeline entered an error state.
- an embodiment of the visual inspection system of the present invention includes:
- the visual collection device 601 is configured to collect an image in the current pipeline
- the visual detection device 602 is configured to acquire an image in a current pipeline collected by the visual system, and determine whether an image feature value of the image is within a correct image feature value threshold;
- the output device 603 is configured to output an abnormality warning when the image feature value of the image is not within the correct image feature value threshold.
- the embodiment of the invention extracts the image feature value of the acquired image, determines whether the image feature value of the image is within the correct image feature value threshold, outputs an abnormal warning or automatically adjusts the correct image feature value threshold, and can timely feedback the acquired image abnormality to avoid The industrial pipeline entered an error state.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
一种视觉检测方法、设备、系统及具有存储功能的装置，该方法包括：获取视觉系统采集的当前流水线中的图像（S1）；提取图像的图像特征值（S2）；判断图像的图像特征值是否在正确图像特征值阈值内（S3）；若否，则输出异常警告或自动调整正确图像特征值阈值（S4）。能够及时反馈获取的图像异常，避免工业流水线进入错误状态。
Description
【技术领域】
本发明涉及视觉检测技术领域，特别涉及一种视觉检测方法、设备、系统及具有存储功能的装置。
【背景技术】
在工业流水线上，对产品的检测一般采用人工抽检的方式，但是由于人工检测劳动强度大，长时间工作易产生疲劳，不能保证检测质量，也无法保证抽检产品的合格率。随着对产品性能、质量稳定性的要求和产量需求的不断提高，人工检测难以适应。而基于视觉系统的检测方式，能够很好地解决以上问题，提高检测效率。
视觉系统在工业流水线中的工作，有时很容易因为相机错误、流水线故障或流水线中受干扰而导致获取图像异常。而在现有的视觉检测技术中，没有针对出现图像异常设立的反馈机制，当出现图像异常时，就会导致整个流水线作业进入错误状态，无法保证工业级应用。
【发明内容】
本发明主要解决的技术问题是提供一种视觉检测方法、设备、系统及具有存储功能的装置，能够及时反馈获取的图像异常，避免工业流水线进入错误状态。
为了解决上述技术问题，本发明采用的一个技术方案是提供一种视觉检测方法，包括：
获取视觉系统采集的当前流水线中的图像；
提取所述图像的图像特征值；
判断所述图像的图像特征值是否在正确特征值阈值内，若否，则输出异常警告或自动调整所述正确图像特征值阈值。
为了解决上述技术问题，本发明采用的另一个技术方案是提供一种视觉检测设备，包括处理器和存储器，所述存储器存储有程序数据，所述处理器加载所述程序数据用于执行上述的方法。
为了解决上述技术问题，本发明采用的另一个技术方案是提供一种具有存储功能的装置，存储有程序数据，所述程序数据适于由处理器加载并执行上述的方法。
为了解决上述技术问题，本发明采用的又一个技术方案是提供一种视觉检测系统，包括：
视觉采集设备，用于采集当前流水线中的图像；
视觉检测设备，用于获取视觉系统采集的当前流水线中的图像，并判断所述图像的图像特征值是否在正确图像特征值阈值内；
输出设备，用于当所述图像的图像特征值不在正确图像特征值阈值内时，输出异常警告。
本发明通过提取采集的图像的图像特征值，判断图像的图像特征值是否在正确图像特征值阈值内而输出异常警告或自动调整正确图像特征值阈值，能够及时反馈获取的图像异常，避免工业流水线进入错误状态。
【附图说明】
图1是本发明视觉检测方法一实施例的流程示意图；
图2是本发明视觉检测方法另一实施例的流程示意图；
图3是本发明视觉检测方法另一实施例的当前采集的图像示意图；
图4是本发明视觉检测设备实施例的结构示意图；
图5是本发明具有存储功能的装置实施例的结构示意图；
图6是本发明视觉检测系统实施例的结构示意图。
【具体实施方式】
参见图1，本发明视觉检测方法一实施例包括：
S1．获取视觉系统采集的当前流水线中的图像；
该视觉系统可以为任意可在当前流水线中采集图像的视觉系统，例如机器视觉系统等。
可选地，通过摄像机、手机等摄像设备进行图像采集。
S2.提取图像的图像特征值；
S3．判断图像的图像特征值是否在正确图像特征值阈值内；
S4．若否，则输出异常警告或自动调整正确图像特征值阈值；
可选地，通过与正确图像的图像特征值阈值进行对比，判断图像是否正确。
可选地，通过声音或闪灯等方式输出警告，例如通过蜂鸣器、报警灯输出警告，或在屏幕上显示报警字样等。
本发明实施例通过提取采集的图像的图像特征值，判断图像的图像特征值是否在正确图像特征值阈值内而输出异常警告或自动调整正确图像特征值阈值，能够及时反馈获取的图像异常，避免工业流水线进入错误状态。
参见图2，本发明视觉检测方法另一实施例包括：
S101．在样本集中存储正确图像样本形成正确图像样本集，并根据形成的正确图像样本集生成正确图像特征值阈值；在样本集中存储异常图像样本形成异常图像样本集，并根据形成的异常图像样本集生成异常图像特征值阈值，异常图像样本集包括对应不同异常类型的异常图像预警类。
具体的，根据流水线上产品的正确标准预先存储正确图像样本，根据可能异常的产品的标准预先存储异常图像样本。
S201．获取视觉系统采集的当前流水线中的图像；
S202．提取图像的图像特征值；
可选地，图像特征值用于判断图像是否异常以及对图像进行目标检测；其中，图像特征值可以是平均亮度、均布平均亮度、被检测目标区周围参考区域的锐度、被检测目标区域的灰度差或局部灰度分布等。
S301．判断图像的图像特征值是否在正确图像特征值阈值内；
可选地，将图像中提取的图像特征值与预先存储的正确图像特征值进行比较，判断提取的图像特征值是否在预先存储的正确图像特征值阈值内。
S401．若图像的图像特征值不在正确图像特征值阈值内，则输出异常警告；
具体的，视觉系统判断图像的图像特征值不在正确图像特征值阈值内，则为异常图像，视觉系统输出异常警告。
S501．若图像的图像特征值不在正确图像特征值阈值内，则判断图像的图像特征值是否在异常图像特征值阈值内；
S601．若图像的图像特征值在异常图像特征值阈值内，则将图像的图像特征值添加到异常图像样本集中。
可选地，将图像中提取的图像特征值与预先存储的异常图像特征值进行比较，判断提取的图像特征值是否在预先存储的异常图像特征值阈值内；若提取的图像特征值在异常图像特征值阈值内，则判断为图像异常，将提取的图像的图像特征值添加到步骤S101的异常图像样本集中，以进行增量学习训练，更新异常图像特征值阈值。
S701．对新的异常图像进行计数；
S702．判断计数是否超过阈值，
S703．若超过阈值，则创建一个新的异常图像预警类，将新的异常图像的图像特征值添加到异常图像样本集中。
可选地，异常图像样本集包括对应不同类型的异常图像预警类，不同的图像预警类对图像的处理方式不一样，例如根据采集的异常图像属于不同的预警类，视觉系统输出异常警告或自动调整正确图像特征值阈值以避免发出异常警告。若视觉系统判断图像的图像特征值既不在正确图像特征值阈值内，同时也不在异常图像特征值阈值内，则判断为新的异常图像，对新的异常图像进行存储并计数，当新的异常图像的数量达到阈值，例如计数达到100次，则由这100张新的异常图像创建一个新的异常图像预警类，将这100张新的异常图像共同抽取的图像的图像特征值添加到步骤S101的异常图像样本集中，以进行增量学习训练，更新异常图像特征值阈值。
S801．对图像进行目标检测；
具体的，若图像的图像特征值在正确图像特征值阈值内，则对图像进行目标检测；对图像进行目标检测所采用的算法与上述判断图像是否异常所采用的算法不同。
S802．判断目标检测是否成功；
S803．若检测成功，则判断为图像正确，通过此检测流程，输出结果；否则将图像的图像特征值添加到异常图像样本集中，以进行增量学习训练，更新异常图像特征值阈值。
可选地，目标检测成功，则判断为图像正确，直接输出检测结果；目标检测失败，则判断为图像异常，将异常的图像的图像特征值返回步骤S601作为学习输入的异常样本添加到异常图像样本集中。
S901．若目标检测成功，则判断为图像正常，将图像的图像特征值添加到正确图像样本集中；
可选地，若提取的图像特征值在正确图像特征值阈值内，并且通过目标检测成功，则判断为图像正确，将提取的图像的图像特征值添加到步骤S101的正确图像样本集中，以进行增量学习训练，更新异常图像特征值阈值。
参见图3，当前采集的图像包括目标检测区域10和参考区域20。
样本集中预先存储正确图像样本集及其正确图像特征值阈值、异常图像样本集及其异常图像特征值阈值，通过对参考区域的值进行学习，更新图像特征值阈值，得到当前参考区域的图像特征值阈值；例如，通过学习得到参考区域的灰度值V1，通过学习获得正确图像特征值阈值为50，当前采集的图像中参考区域20的灰度值为V2，当V2＞V1+50，超过正确图像特征值阈值，则判断参考区域20特征值异常，输出异常图像警告。
可选的，正确图像特征值阈值也可以通过人为设定，在这里不做限制。
本发明实施例通过对采集的图像进行检测和分析，判断图像的图像特征值是否在正确图像特征值阈值或异常图像特征值阈值内而输出异常警告，并将图像的图像特征值作为学习输入的样本添加到样本集，更新图像特征值阈值，能够及时反馈获取的图像异常，避免工业流水线进入错误状态，并不断更新完善样本集，使分析结果更加准确，提高检测效率。
参见图4，本发明视觉检测设备40实施例包括处理器401和存储器402，存储器402存储有程序数据，处理器401加载程序数据用于执行上述的视觉检测方法。
本发明实施例通过提取采集的图像的图像特征值，判断图像的图像特征值是否在正确图像特征值阈值内而输出异常警告或自动调整正确图像特征值阈值，能够及时反馈获取的图像异常，避免工业流水线进入错误状态。
参见图5，本发明具有存储功能的装置50实施例，存储有程序数据501，程序数据501适于由处理器加载并执行上述的视觉检测方法。
本发明实施例通过提取采集的图像的图像特征值，判断图像的图像特征值是否在正确图像特征值阈值内而输出异常警告或自动调整正确图像特征值阈值，能够及时反馈获取的图像异常，避免工业流水线进入错误状态。
参见图6，本发明视觉检测系统实施例包括：
视觉采集设备601，用于采集当前流水线中的图像；
视觉检测设备602，用于获取视觉系统采集的当前流水线中的图像，并判断图像的图像特征值是否在正确图像特征值阈值内；
输出设备603，用于当图像的图像特征值不在正确图像特征值阈值内时，输出异常警告。
本实施例的具体实施过程参见上述视觉检测方法实施例，在此不再赘述。
本发明实施例通过提取采集的图像的图像特征值，判断图像的图像特征值是否在正确图像特征值阈值内而输出异常警告或自动调整正确图像特征值阈值，能够及时反馈获取的图像异常，避免工业流水线进入错误状态。
以上所述仅为本发明的实施方式，并非因此限制本发明的专利范围，凡是利用本发明说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本发明的专利保护范围内。
Claims (11)
- 一种视觉检测方法，其特征在于，包括：获取视觉系统采集的当前流水线中的图像；提取所述图像的图像特征值；判断所述图像的图像特征值是否在正确图像特征值阈值内，若否，则输出异常警告或自动调整所述正确图像特征值阈值。
- 根据权利要求1所述的方法，其特征在于，若所述图像的图像特征值在正确图像特征值阈值内，则对所述图像进行目标检测。
- 根据权利要求1或2所述的方法，其特征在于，所述获取视觉系统采集的当前流水线中的图像之前，还包括：存储正确图像样本形成正确图像样本集，并根据所述形成的正确图像样本集生成正确图像特征值阈值；存储异常图像样本形成异常图像样本集，并根据所述形成的异常图像样本集生成异常图像特征值阈值，所述异常图像样本集包括对应不同异常类型的异常图像预警类。
- 根据权利要求3所述的方法，其特征在于，若所述图像的图像特征值在异常图像特征值阈值内，则将所述图像的图像特征值添加到所述异常图像样本集中，以进行增量学习训练，更新所述异常图像特征值阈值。
- 根据权利要求3所述的方法，其特征在于，若所述图像的图像特征值既不在正常图像特征值阈值内也不在异常图像特征值阈值内，则创建新的异常图像预警类，将所述图像的图像特征值添加到所述异常图像样本集中。
- 根据权利要求5所述的方法，其特征在于，所述创建新的异常图像预警类之前，包括：对图像特征值既不在正常图像特征值阈值内也不在异常图像特征值阈值内的所述图像进行计数；判断所述计数是否超过阈值，若超过阈值，才执行所述创建新的异常图像预警类的步骤。
- 根据权利要求2所述的方法，其特征在于，所述对所述图像进行目标检测包括：若检测成功，则判断所述图像正确，让所述图像中的目标通过此检测流程，输出结果；否则将所述图像的图像特征值添加到所述异常图像样本集中，以进行增量学习训练，更新所述异常图像特征值阈值。
- 根据权利要求2所述的方法，其特征在于，所述对所述图像进行目标检测还包括：若目标检测成功，则判断为所述图像正确，并将所述图像的图像特征值添加到所述正确图像样本集中，以进行增量学习训练，更新所述正确图像特征值阈值。
- 一种视觉检测设备，包括处理器和存储器，所述存储器存储有程序数据，所述处理器加载所述程序数据用于执行如权利要求1-8任一项所述的方法。
- 一种具有存储功能的装置，存储有程序数据，所述程序数据适于由处理器加载并执行权利要求1-8任一项所述的方法。
- 一种视觉检测系统，包括：视觉采集设备，用于采集当前流水线中的图像；视觉检测设备，用于获取视觉系统采集的当前流水线中的图像，并判断所述图像的图像特征值是否在正确图像特征值阈值内。输出设备，用于当所述图像的图像特征值不在正确图像特征值阈值内时，输出异常警告。
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201780035663.9A CN109564632A (zh) | 2017-05-24 | 2017-05-24 | 一种视觉检测方法、设备、系统及具有存储功能的装置 |
| PCT/CN2017/085708 WO2018214054A1 (zh) | 2017-05-24 | 2017-05-24 | 一种视觉检测方法、设备、系统及具有存储功能的装置 |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2017/085708 WO2018214054A1 (zh) | 2017-05-24 | 2017-05-24 | 一种视觉检测方法、设备、系统及具有存储功能的装置 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2018214054A1 true WO2018214054A1 (zh) | 2018-11-29 |
Family
ID=64396167
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2017/085708 WO2018214054A1 (zh) | 2017-05-24 | 2017-05-24 | 一种视觉检测方法、设备、系统及具有存储功能的装置 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN109564632A (zh) |
| WO (1) | WO2018214054A1 (zh) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112445240A (zh) * | 2020-11-26 | 2021-03-05 | 广东电网有限责任公司 | 一种自动巡线方法和自动巡线无人机 |
| CN113552123A (zh) * | 2020-04-17 | 2021-10-26 | 华为技术有限公司 | 视觉检测方法和视觉检测装置 |
| CN115414977A (zh) * | 2021-05-13 | 2022-12-02 | 南京智欧智能技术研究院有限公司 | 一种智能移液系统及移液方法 |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110782431B (zh) * | 2019-10-08 | 2022-03-22 | 浙江大学 | 一种基于深度学习的高压电线结冰区域检测方法 |
| CN111524107B (zh) * | 2020-04-15 | 2023-02-28 | Oppo（重庆）智能科技有限公司 | 缺陷检测方法、缺陷检测装置及计算机可读存储介质 |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101169826A (zh) * | 2007-11-29 | 2008-04-30 | 上海交通大学 | 基于机器视觉的排辊成形焊管质量在线监控系统 |
| CN102601131A (zh) * | 2012-03-09 | 2012-07-25 | 湖南镭目科技有限公司 | 一种钢坯表面质量在线检测装置 |
| CN102706881A (zh) * | 2012-03-19 | 2012-10-03 | 天津工业大学 | 基于机器视觉的布匹瑕疵检测方法 |
| US20130022236A1 (en) * | 2011-07-20 | 2013-01-24 | Altek Corporation | Apparatus Capable of Detecting Location of Object Contained in Image Data and Detection Method Thereof |
| CN104537359A (zh) * | 2015-01-15 | 2015-04-22 | 浙江宇视科技有限公司 | 一种车辆目标检测方法及装置 |
| CN106338520A (zh) * | 2016-09-18 | 2017-01-18 | 南京林业大学 | 一种表板为拼板的多层实木复合地板表面缺陷的识别方法 |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104866869B (zh) * | 2015-05-29 | 2018-12-14 | 武汉大学 | 基于分布差异与增量学习的时序sar图像分类方法 |
| CN105205503B (zh) * | 2015-08-28 | 2018-09-18 | 重庆恢恢信息技术有限公司 | 基于众包主动学习用于检测异常图片的方法 |
| CN105590401B (zh) * | 2015-12-15 | 2019-08-20 | 天维尔信息科技股份有限公司 | 基于视频图像的预警联动方法及系统 |
| CN105447530A (zh) * | 2016-01-05 | 2016-03-30 | 国网四川省电力公司检修公司 | 基于图像识别技术的输电线路隐患及故障检测方法 |
| CN106097346B (zh) * | 2016-06-13 | 2019-08-27 | 中国科学技术大学 | 一种自学习的视频火灾探测方法 |
| CN106446807A (zh) * | 2016-09-09 | 2017-02-22 | 成都通甲优博科技有限责任公司 | 一种井盖失窃检测方法 |
-
2017
- 2017-05-24 CN CN201780035663.9A patent/CN109564632A/zh active Pending
- 2017-05-24 WO PCT/CN2017/085708 patent/WO2018214054A1/zh active Application Filing
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101169826A (zh) * | 2007-11-29 | 2008-04-30 | 上海交通大学 | 基于机器视觉的排辊成形焊管质量在线监控系统 |
| US20130022236A1 (en) * | 2011-07-20 | 2013-01-24 | Altek Corporation | Apparatus Capable of Detecting Location of Object Contained in Image Data and Detection Method Thereof |
| CN102601131A (zh) * | 2012-03-09 | 2012-07-25 | 湖南镭目科技有限公司 | 一种钢坯表面质量在线检测装置 |
| CN102706881A (zh) * | 2012-03-19 | 2012-10-03 | 天津工业大学 | 基于机器视觉的布匹瑕疵检测方法 |
| CN104537359A (zh) * | 2015-01-15 | 2015-04-22 | 浙江宇视科技有限公司 | 一种车辆目标检测方法及装置 |
| CN106338520A (zh) * | 2016-09-18 | 2017-01-18 | 南京林业大学 | 一种表板为拼板的多层实木复合地板表面缺陷的识别方法 |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113552123A (zh) * | 2020-04-17 | 2021-10-26 | 华为技术有限公司 | 视觉检测方法和视觉检测装置 |
| CN112445240A (zh) * | 2020-11-26 | 2021-03-05 | 广东电网有限责任公司 | 一种自动巡线方法和自动巡线无人机 |
| CN115414977A (zh) * | 2021-05-13 | 2022-12-02 | 南京智欧智能技术研究院有限公司 | 一种智能移液系统及移液方法 |
| CN115414977B (zh) * | 2021-05-13 | 2023-10-10 | 南京智欧智能技术研究院有限公司 | 一种智能移液系统及移液方法 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109564632A (zh) | 2019-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2018214054A1 (zh) | 一种视觉检测方法、设备、系统及具有存储功能的装置 | |
| WO2018101722A2 (ko) | 머신 러닝 기반 반도체 제조 수율 예측 시스템 및 방법 | |
| WO2017191872A1 (ko) | 플랜트 이상 감지 방법 및 시스템 | |
| WO2019107614A1 (ko) | 제조 공정에서 딥러닝을 활용한 머신 비전 기반 품질검사 방법 및 시스템 | |
| WO2019088462A1 (ko) | 혈압 추정 모델 생성 시스템 및 방법과 혈압 추정 시스템 및 방법 | |
| WO2016171341A1 (ko) | 클라우드 기반 병리 분석 시스템 및 방법 | |
| CN107862338A (zh) | 基于双检验法的海洋环境监测数据质量管理方法及系统 | |
| US7872584B2 (en) | Analyzing smoke or other emissions with pattern recognition | |
| WO2020130306A1 (ko) | 검증용 주석 처리 작업을 이용한 실전용 주석 처리 작업의 검증 방법 및 장치 | |
| WO2021020866A1 (ko) | 원격 모니터링을 위한 영상 분석 시스템 및 방법 | |
| WO2019132476A1 (ko) | 고로 조업 상황 평가 시스템 및 방법 | |
| WO2023128669A1 (ko) | 부분방전 모니터링 시스템 및 부분방전 모니터링 방법 | |
| WO2022265197A1 (ko) | 인공지능에 기반하여 내시경 영상을 분석하기 위한 방법 및 장치 | |
| WO2015196878A1 (zh) | 一种电视虚拟触控方法及系统 | |
| WO2018164331A1 (ko) | 변압기 수소 가스 감시 시스템, 장치 및 방법 | |
| WO2019240330A1 (ko) | 영상 기반 힘 예측 시스템 및 그 방법 | |
| WO2021006501A1 (ko) | 스마트 재봉작업 측정 방법 및 이를 수행하기 위한 시스템 | |
| WO2023158068A1 (ko) | 객체검출률 향상을 위한 학습시스템 및 그 방법 | |
| WO2018006443A1 (zh) | 一种显示屏残影检测系统及其方法 | |
| WO2017010612A1 (ko) | 의료 영상 분석 기반의 병리 진단 예측 시스템 및 방법 | |
| WO2023090825A1 (ko) | Ai 모델 드리프트 모니터링 장치 및 방법 | |
| WO2021261905A1 (ko) | 영상 분석 기반 작업 동작 인식 및 생산량 측정 장치와 그 방법 | |
| WO2022255518A1 (ko) | 딥러닝 신경망 모델을 이용한 검사대상 패널의 결함 판정장치 | |
| WO2022107975A1 (ko) | 시계열 데이터의 변화 추세를 고려한 유사도 분석 방법 | |
| JP6898382B2 (ja) | 射出成形機稼動管理に対する支援方法 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 17911079 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 17911079 Country of ref document: EP Kind code of ref document: A1 |