Method of generating image data for dangerous object detection and imaging system using the same

1 . An image data generating method comprising:

performing operations of emitting electromagnetic waves to an object and receiving scan signals scanning the object, by using an electromagnetic wave transceiver;

generating first image data of a low resolution based on the scan signals;

generating second image data of a high resolution associated with a sub-region of the first image data when at least one of a suspicious object and a target object candidate is detected in the sub-region; and

detecting, from the second image data, a final target object different from both the suspicious object and the target object candidate.

2 . The image data generating method of claim 1 , wherein the first image data are generated by using one or all of a fast Fourier transform technique and a low-resolution back-projection technique, and wherein the second image data are generated by using a high-resolution back-projection technique.

3 . The image data generating method of claim 2 , further comprises:

generating third image data based on the first image data and the second image data when the final target object is detected.

4 . The image data generating method of claim 3 , wherein the generating of the second image data includes:

determining whether the target object candidate is detected from the first image data, by using a first deep learning model, and

wherein the generating of the third image data includes:

determining whether the final target object is detected from the second image data, by using a second deep learning model.

5 . The image data generating method of claim 4 , wherein dangerous objects are classified based on a hierarchical structure so as to be included in an upper category and a lower category, wherein the target object candidate is one of first detailed items included in the upper category, and wherein the final target object is one of second detailed items included in the lower category.

6 . The image data generating method of claim 5 , wherein the first deep learning model detects one of the first detailed items as the target object candidate, and wherein the second deep learning model detects one of the second detailed items as the final target object.

7 . The image data generating method of claim 3 , wherein the generating of the third image data includes:

composing the first image data and the second image data.

8 . The image data generating method of claim 7 , wherein the first image data are generated with respect to an entire scan region where the object is scanned, and wherein the second image data are generated only with respect to the sub-region of the first image data.

9 . The image data generating method of claim 8 , wherein the third image data are generated by replacing the sub-region of the first image data with the second image data.

10 . The image data generating method of claim 1 , wherein the generating of the first image data includes:

generating preprocessed data by performing noise cancellation, error correction, or data alignment with respect to the scan signals; and

generating the first image data by processing the preprocessed data by using one or all of a fast Fourier transform technique and a low-resolution back-projection technique.

11 . The image data generating method of claim 1 , wherein the electromagnetic wave transceiver includes:

a transmitter configured to emit the terahertz wave to a scan region; and

a receiver configured to receive the scan signals when the object is present in the scan region.

12 . The image data generating method of claim 3 , wherein each of the first image data, the second image data, and the third image data is three-dimensional (3D) image data.

13 . An imaging system comprising:

an electromagnetic wave transceiver configured to emit electromagnetic waves to an object and to receive scan signals scanning the object; and

an image data generating device configured to generate synthetic image data based on the scan signals,

wherein the image data generating device is configured to:

generate first image data of a low resolution based on the scan signals;

generate second image data of a high resolution associated with a sub-region of the first image data when at least one of a suspicious object and a target object candidate is detected in the sub-region; and

detect a final target object different from both the suspicious object and the target object candidate.

14 . The imaging system of claim 13 , wherein the image data generating device includes:

a first image data generator configured to generate the first image data by applying one or all of a fast Fourier transform technique and a low-resolution back-projection technique to the scan signals;

a second image data generator configured to detect a target object candidate from the first image data and to generate the second image data by applying a high-resolution back-projection technique to the sub-region of the first image data; and

a third image data generator configured to detect a final target object from the second image data and to generate a third image data based on the first image data and the second image data.

15 . The imaging system of claim 14 , wherein the image data generating device further includes:

a first deep learning model configured to detect the target object candidate; and

a second deep learning model configured to detect the final target object.

16 . The imaging system of claim 15 , wherein dangerous objects are classified based on a hierarchical structure so as to be included in an upper category and a lower category, wherein the target object candidate is one of first detailed items included in the upper category, and wherein the final target object is one of second detailed items included in the lower category.

17 . The imaging system of claim 14 , wherein the third image data generator is configured to generate the third image data by replacing the sub-region of the first image data with the second image data.

18 . The imaging system of claim 13 , wherein the electromagnetic wave transceiver includes:

a transmitter configured to emit terahertz waves to a scan region; and

a receiver spaced apart from the transmitter and configured to receive the scan signals when the object is present in the scan region.