Inspection devices and methods for inspecting a container

1. An inspection device, comprising:

a scanning device comprising a ray source and a detector, and configured to perform X transmission scanning on an inspected container to obtain a scanned image;

a data processing device configured to:

process the scanned image to obtain a region of interest;

calculate features of texture units included in the region of interest;

form local descriptions of the texture units using the features of the texture units;

calculate distinction of each local point from a local description of each of the texture units so as to obtain a local distinct map of the region of interest;

determine at least one candidate regions for the article which is secretly carried by using the local distinct map;

wherein the data processing device is further configured to

determine distinction values of the candidate regions and a number of adjacent candidate points of the candidate regions, respectively; and

determine that there is an article secretly carried in the candidate region for which the distinction value is greater than a first threshold and the number of the adjacent candidate points is greater than a second threshold.

2. The inspection device according to claim 1 , wherein the data processing device is configured to:

reduce complexity of the local description using an integrate image method; and

calculate the distinction of the local description using a sliding window with a fixed step size.

3. The inspection device according to claim 1 , wherein the data processing device is configured to calculate similarities between any local description and local descriptions which is adjacent to the local description and is not overlapped with the local description, and determine a maximum value of the similarities as distinction of a center of the local description.

4. The inspection device according to claim 1 , wherein the data processing device is configured to calculate Euclidean distances between any local description and local descriptions which is adjacent to the local description and is not overlapped with the local description, and determine a minimum value of the Euclidean distances as distinction of a center of the local description.

5. The inspection device according to claim 1 , wherein the data processing device is configured to calculate local distinct maps on different scales, overlapping the local distinct maps on different scales, and then determine whether there is an article which is secretly carried in the inspected container using the overlapped local distinct maps.

6. A method for inspecting a container, comprising steps of:

performing X-ray scanning on the inspected container to obtain a scanned image;

processing the scanned image to obtain a region of interest;

calculating features of texture units included in the region of interest;

forming local descriptions of the texture units based on the features of the texture units;

calculating distinction of each local point from a local description of each of the texture units so as to obtain a local distinct map of the region of interest; and

determining at least one candidate regions for the article which is secretly carried by using the local distinct map; and

determining distinction values of the candidate regions and a number of adjacent candidate points of the candidate regions, respectively; and

wherein that the method further comprising

determining that there is an article secretly carried in the candidate region for which the distinction value is greater than a first threshold and the number of the adjacent candidate points is greater than a second threshold.

7. The method according to claim 6 , wherein the features of the texture units are represented as a multi-dimensional vector, and the method further comprises a step of:

reducing a dimension of the multi-dimensional vector to 1 by means of a dictionary description.

8. The method according to claim 6 , wherein the step of calculating (S 330 ) distinction of each local point from a local description of each of the texture units comprises:

reducing complexity of the local description using an integrate image method; and

calculating the distinction of the local description using a sliding window with a fixed step size.

9. The method according to claim 6 , further comprising:

calculating similarities between any local description and local descriptions which is adjacent to the local description and is not overlapped with the local description, and determining a maximum value of the similarities as distinction of a center of the local description.

10. The method according to claim 6 , further comprising:

calculating Euclidean distances between any local description and local descriptions which is adjacent to the local description and is not overlapped with the local description, and determining a minimum value of the Euclidean distances as distinction of a center of the local description.

11. The method according to claim 6 , wherein the step of determining whether there is an article which is secretly carried in the inspected container using the local distinct map comprises:

calculating local distinct maps on different scales, overlapping the local distinct maps on different scales, and then determining whether there is an article which is secretly carried in the inspected container using the overlapped local distinct maps.