Inspection methods and systems

Patent metadata

Field	Value
Publication number	US-10509979-B2
Application number	US-201916255738-A
Country	US
Kind code	B2
Filing date	Jan 23, 2019
Priority date	Dec 30, 2014
Publication date	Dec 17, 2019
Grant date	Dec 17, 2019

Abstract

An inspection method and system for inspecting whether there is any liquor in goods is provided. The method includes: acquiring a radiation image of goods being inspected; processing on the radiation image to obtain an ROI; inspecting on the ROI using a liquor goods inspection model to determine if the ROI of the radiation image contains liquor goods. The above solution performs liquor inspection on scanned images of goods, especially containers, so as to intelligently assist the image inspectors.

First claim

What is claimed is:

1. An inspection method comprising:
acquiring a radiation image of goods being inspected;
processing on the radiation image to obtain an ROI;
inspecting on the ROI using a liquor goods inspection model to determine if the ROI of the radiation image contains liquor goods, wherein the liquor goods inspection model is established based on a weighted sum of shape information and texture information of a feature in samples, and the weight is in association with a regional conditional entropy of the feature.

2. The inspection method according to claim 1 , wherein the step of inspecting on the ROI using a liquor goods inspection model comprising:
extracting shape information and texture information of a local target from the ROI;
classifying the shape information and the texture information of the local target from the ROI using the liquor goods inspection model established based on a weighted sum of shape information and texture information of a feature in the local target, so as to determine if the ROI of the radiation image contains liquor goods, the weight is in association with a regional conditional entropy of the feature.

3. The inspection method according to claim 1 , wherein the ROI is inspected in multiple-scales using the liquor goods inspection model.

4. The inspection method according to claim 1 , wherein the step of processing on the radiation image comprising:
detecting on air regions and impenetrable regions in the radiation image;
excluding the air regions and the impenetrable regions from the ROIs.

5. The inspection method according to claim 1 , further comprising a step of:
training the liquor goods inspection model by way of manually labeling using scanned images of known goods categories where liquor goods are contained and scanned images of goods that is similar to liquors but not liquors.

6. The inspection method according to claim 5 , wherein the manually labeling comprises: labeling positions and placing postures of the liquor goods in the image.

7. The inspection method according to claim 5 , wherein the liquor goods inspection model is established with respect to different placing postures of liquor goods.

8. The inspection method according to claim 1 , further comprising a step of:
for those liquor goods that are not detected using the liquor goods inspection model and those goods that are detected using the liquor goods inspection model as liquor goods but turns out to be non-liquor goods, re-training the liquor goods inspection model by manually labeling.

9. An inspection system comprising:
a scanning imaging system configured to scan goods being inspected so as to acquire a radiation image of the goods being inspected;
a data processing apparatus configured to process on the radiation image to obtain an ROI, and to inspect on the ROI using a liquor goods inspection model to determine if the ROI of the radiation image contains liquor goods, wherein the liquor goods inspection model is established based on a weighted sum of shape information and texture information of a feature in samples, the weight is in association with a regional conditional entropy of the feature.

10. The inspection system according to claim 9 , wherein the data processing apparatus is configured to extract shape information and texture information of a local target from the ROI, and to classify the shape information and the texture information of the local target from the ROI using the liquor goods inspection model established based on a weighted sum of shape information and texture information of a feature in the local target so as to determine if the ROI of the radiation image contains liquor goods.

Assignees

• Nuctech Co Ltd

Inventors

• Chen Zhiqiang
• Zhang Li
• Zhao Ziran
• Liu Yaohong
• Zhang Duokun
• Gu Jianping
• Li Qiang
• Zhang Jian

Classifications

• G01N23/046

  using tomography, e.g. computed tomography [CT] · CPC title

• G06K9/3241 Primary

  Physics · mapped topic

• G01V5/005

  Physics · mapped topic

• G01V5/0016

  Physics · mapped topic

• G01V5/22 Primary

  Active interrogation, i.e. by irradiating objects or goods using external radiation sources, e.g. using gamma rays or cosmic rays · CPC title