An image dataset extension method for terahertz imaging security inspection system

Abstract

The application provides an image dataset expansion method for a terahertz imaging security check system, which can solve the problem of poor continuity and unsmooth video display of millimeter wave / terahertz video images caused by the insufficient actual available data sample amount due to the influence of factors such as the limitation of system imaging equipment and long imaging time in the current millimeter wave / terahertz imaging security check process; a single-frame original terahertz imaging image of a measured target person is collected by a terahertz imaging security check device; each collected original terahertz imaging image is subjected to image expansion processing and an expansion new image corresponding to each original terahertz imaging image is obtained, especially, a semantic segmentation technology is applied to obtain the outer contour image of the measured target person in each original terahertz imaging image, and the position of the measured target person in the original terahertz imaging is judged according to the outer contour image, then the imaging image is subjected to position bias processing according to the position, and the image after the bias processing is added to the current frame as a new image.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of security inspection, in particular to the field of millimeter wave / terahertz imaging security inspection, and specifically to an image dataset expansion method for a terahertz imaging security inspection system. BACKGROUND

[0002] Millimeter wave / terahertz security inspection technology is a new type of human body security inspection technology emerging in recent years. It uses millimeter wave or terahertz wave for non-contact security inspection of human bodies. It has the advantages of being harmless to human bodies, fast security inspection speed, and high security inspection accuracy, and is gradually replacing traditional metal detection, X-ray irradiation, and other security inspection forms. Existing millimeter wave / terahertz security inspection equipment mostly uses deep learning technology to continuously improve detection performance by training collected data samples. However, due to the limitations of current millimeter wave / terahertz system imaging equipment, long imaging time, and other factors, the actual available data sample size is insufficient in specific applications. There are problems of poor real-time performance of millimeter wave images and low frame rate of image acquisition. This leads to poor continuity of millimeter wave video images and unsmooth video display, and still cannot achieve good practical application results. SUMMARY

[0003] To solve the above problems, the present application provides an image dataset expansion method for a terahertz imaging security inspection system, which can solve the problem of poor continuity of millimeter wave / terahertz video images and unsmooth video display caused by insufficient actual available data sample size in the current millimeter wave / terahertz imaging security inspection process due to the limitations of system imaging equipment, long imaging time, and other factors.

[0004] The technical scheme is an image dataset expansion method for a terahertz imaging security inspection system, characterized in that it comprises the following steps:

[0005] S100, a terahertz imaging security inspection device collects an original terahertz imaging image of a target person to be measured;

[0006] S200, image expansion processing is performed on each frame of the collected original terahertz imaging image to obtain a corresponding expanded new image; Figure One

[0007] S300, each frame of the expanded new image is inserted into the corresponding original terahertz imaging image;

[0008] In step S200, the image expansion processing of each frame of the original terahertz imaging image comprises the following steps in turn:

[0009] S210, the imaging profile of the target person to be measured in the current frame of the original terahertz imaging image is preprocessed, and the preprocessed image is taken as a to-be-processed image;​

[0010] S220, judging geometric position information of the imaging profile of the measured target person in the current image to be processed;

[0011] S230, predicting the offset direction of the measured target person in the extended new image according to the geometric position information obtained in step S220;

[0012] S240, performing overall offset processing on the imaging profile of the measured target person in the image to be processed according to the offset direction predicted in step S230, to obtain a bias image based on the current single-frame original terahertz imaging image;

[0013] S250, performing intersection processing on the bias image obtained in step S240 and the current frame original terahertz imaging image, and then performing scaling processing, to obtain an extended new image based on the current single-frame original terahertz imaging image.

[0014] Further, the pre-processing of the imaging profile of the measured target person in the current frame original terahertz imaging image in step S210 is to obtain the complete outer contour of the imaging of the measured target person in the current frame original terahertz imaging image by using semantic segmentation technology.

[0015] Further, in step S220, an outer rectangle of the outer contour edge of the human body in the image to be processed obtained in step S210 is obtained, and the center point P position of the outer rectangle is determined, the image center point Q position of the image to be processed is obtained, and the geometric position of the imaging profile of the measured target person in the current image to be processed relative to the image center point Q of the image to be processed is determined according to the positions of the center point P of the outer rectangle and the image center point Q of the image to be processed.

[0016] Further, a rectangular coordinate system is established on the image to be processed with the image center point Q as the origin, and the position of the center point P of the outer rectangle in the rectangular coordinate system is determined.

[0017] Further, the rectangular coordinate system divides the image to be processed into four regions, namely the upper left region, the upper right region, the lower left region, and the lower right region; in step S230, if the center point P of the outer rectangle is located in the upper left region, it is predicted that the imaging of the measured target person in the extended new image moves left in the horizontal direction and moves up in the vertical direction of the rectangular coordinate system; if the center point P of the outer rectangle is located in the upper right region, it is predicted that the imaging of the measured target person in the extended new image moves right in the horizontal direction and moves up in the vertical direction of the rectangular coordinate system; if the center point P of the outer rectangle is located in the lower left region, it is predicted that the imaging of the measured target person in the extended new image moves left in the horizontal direction and moves down in the vertical direction of the rectangular coordinate system; and if the center point P of the outer rectangle is located in the lower right region, it is predicted that the imaging of the measured target person in the extended new image moves right in the horizontal direction and moves down in the vertical direction of the rectangular coordinate system.

[0018] Further, in step S240, the whole profile of the measured target person in the image to be processed is offset by 1-2 pixels.

[0019] Further, in step S250, the intersection processing is performed by aligning the offset image with the corresponding current single-frame original terahertz imaging data image, cutting off the non-overlapping part in the offset image, and then scaling the remaining part to the size of the single-frame original terahertz imaging data image to form the expanded new image.

[0020] The application also provides a computer device, which comprises a processor, a memory and a program; the program is stored in the memory, and the processor calls the program stored in the memory to execute the image data set expansion method for a terahertz imaging security inspection system according to claims 1-5.

[0021] The application also provides a computer readable storage medium, which is used for storing a program for executing the image data set expansion method for a terahertz imaging security inspection system according to claims 1-5.

[0022] The application has the advantages that: the original terahertz imaging image is analyzed, the semantic segmentation technology is applied to obtain the outer contour image of the measured target person in each original terahertz imaging image, the geometric position of the measured target person in the original terahertz imaging is determined, the position of the imaging image is offset according to the position, the offset image is added to the current frame as a new image, the gap image missed in the original image frame rate is made up, the progress of the measured target person in the security inspection is supplemented, the detection image is natural and complete, the sample number for deep learning training is expanded, the security inspection accuracy is improved, and good practical application effect is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 The flowchart of the image data set expansion method for a terahertz imaging security inspection system is shown in the application;

[0024] Figure 2 The flowchart of step S200 in the method of the application is shown in the application;

[0025] Figure 3 The original terahertz imaging image of the front of the measured target person collected by the terahertz imaging security inspection device through step S100 in the embodiment of the application is shown in the application;

[0026] Figure 4The original terahertz imaging image of the back of the target person collected by the terahertz imaging security inspection device in step S100 in the embodiment of the present application;

[0027] Figure 5 The processing image formed by the outer contour of the human body of the target person extracted from the 7th frame of the original terahertz imaging image in the embodiment of the present application; Figure 4

[0028] Figure 6 The schematic diagram for determining the human body position of the target person in the processing image in the embodiment of the present application;

[0029] Figure 7 The terahertz imaging image of the front of the target person obtained after the method of the present application is used in the embodiment;

[0030] Figure 8 The terahertz imaging image of the back of the target person obtained after the method of the present application is used in the embodiment;

[0031] Figure 9 The internal structure diagram of the computer device in one embodiment of the present application. DETAILED DESCRIPTION

[0032] See Figure 1 and Figure 2 , the present application is a kind of for the image data set extension method of terahertz imaging security inspection system, it includes the following steps,

[0033] S100, terahertz imaging security inspection equipment collects the single frame original terahertz imaging image of target person;In the present embodiment, terahertz imaging security inspection equipment collects 16 frames of single frame original terahertz imaging image of the front and back of a target person respectively, see Figure 3 and Figure 4 ;

[0034] S200, the image extension processing of each frame original terahertz imaging image collected is carried out and the corresponding expansion new image of each frame original terahertz imaging Figure One is obtained;

[0035] Next, the 7th frame of the back single frame original terahertz imaging image in the embodiment is taken as an example to specifically describe the implementation process of step S200:

[0036] S210, the imaging contour of the target person in the current frame original terahertz imaging image is preprocessed, specifically, the outer contour of the human body of the target person is extracted from the 7th frame of the back imaging image by applying semantic segmentation technology, and the processing image as shown in Figure 5 is formed;

[0037] ​S220, judging the geometric position information of the imaging profile of the measured target person in the current image to be processed; acquiring Figure 5 the circumscribed rectangle of the outer contour of the human body, and determining the position of the center point P of the circumscribed rectangle, acquiring the position of the image center point Q of the image to be processed, see Figure 6 , according to the positions of the center point P of the circumscribed rectangle and the image center point Q of the image to be processed, judging the geometric position of the imaging profile of the measured target person in the current image to be processed relative to the image center point Q of the image to be processed;

[0038] As a preferred technical scheme of the present application, a rectangular coordinate system is established on the image to be processed with the image center point Q as the origin, and the position of the center point P of the circumscribed rectangle in the rectangular coordinate system is determined.

[0039] As a further preferred technical scheme of the present application, the rectangular coordinate system divides the image to be processed into four regions, namely the upper left region, the upper right region, the lower left region and the lower right region, and the center point P of the circumscribed rectangle is determined to be located in one of the four regions.

[0040] In step S230, if the center point P of the circumscribed rectangle is located in the upper left region, it is predicted that the imaging of the measured target person in the expanded new image moves left in the horizontal direction and moves up in the vertical direction of the rectangular coordinate system; if the center point P of the circumscribed rectangle is located in the upper right region, it is predicted that the imaging of the measured target person in the expanded new image moves right in the horizontal direction and moves up in the vertical direction of the rectangular coordinate system; if the center point P of the circumscribed rectangle is located in the lower left region, it is predicted that the imaging of the measured target person in the expanded new image moves left in the horizontal direction and moves down in the vertical direction of the rectangular coordinate system; if the center point P of the circumscribed rectangle is located in the lower right region, it is predicted that the imaging of the measured target person in the expanded new image moves right in the horizontal direction and moves down in the vertical direction of the rectangular coordinate system; the Figure 6 In this embodiment, the center point P of the circumscribed rectangle is located in the lower left region, so it is predicted that the imaging of the measured target person in the expanded new image moves left in the horizontal direction and moves down in the vertical direction of the rectangular coordinate system.

[0041] S240, according to the offset direction predicted in step S230, the profile of the measured target person in the image to be processed is shifted left by 1-2 pixels in the horizontal direction and shifted down by 1-2 pixels in the vertical direction of the rectangular coordinate system, to obtain a bias image based on the current single-frame original terahertz imaging image; in this embodiment, the outer contour of the human body in the image to be processed is shifted left by 1 pixel in the horizontal direction and shifted down by 1 pixel in the vertical direction. Figure 5

[0042] S250, after the intersection processing of the bias image obtained in step S240 and the current frame original terahertz imaging image, the scaling processing is performed, to obtain an expanded new image based on the current single-frame original terahertz imaging image.

[0043] ​The intersection processing is that the bias image is superimposed and compared with the corresponding current single-frame original terahertz imaging data image, the non-overlapping part in the bias image is cut off, and the remaining part image is zoomed to the size of the single-frame original terahertz imaging data image to form an extended new image.

[0044] S300, each frame of the extended new image is inserted into the original terahertz imaging image corresponding thereto.

[0045] After the method is adopted, the front terahertz imaging image and the back terahertz imaging image of the measured target person in the embodiment are as follows: Figure 7 and Figure 8 , Figure 7 and Figure 8 The imaging images with odd serial numbers in the above are original imaging images, and the imaging images with even serial numbers are new images processed by the method, which are essentially inserting a new image processed based on the previous original imaging image between the original adjacent two single-frame imaging images, thereby compensating for the missing gap image at a certain degree and supplementing the progress of the measured target person in security inspection, so that the detection image is natural and complete, the number of samples available for deep learning training is more sufficient, the security inspection accuracy is improved, and good practical application effect is achieved.

[0046] In the embodiment of the application, a computer device is also provided, which comprises a processor, a memory and a program.

[0047] The program is stored in the memory, and the processor calls the program stored in the memory to execute the real-time fusion type millimeter wave imaging frame increasing method.

[0048] The computer device can be a terminal, and its internal structure diagram can be as shown in Figure 8 The computer device comprises a processor, a memory, a network interface, a display screen and an input device connected through a bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to implement the interception policy derivation method for credit fraud prevention. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0049] The memory can be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), and the like. Among them, the memory is used to store a program, and the processor executes the program after receiving an execution instruction.

[0050] The processor can be an integrated circuit chip with processing capability. The processor mentioned above can be a general processor, including a central processing unit (CPU), a network processor (NP), and the like. The processor can also be other general processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. The general processor can be a microprocessor or the processor can also be any conventional processor and the like. The general processor can implement or execute the disclosed methods, steps and logic block diagrams in the embodiments of the present application. The general processor can be a microprocessor or any conventional processor and the like.

[0051] Those skilled in the art can understand that, Figure 9 Those skilled in the art can understand that,

[0052] In the embodiments of the present application, a computer readable storage medium is also provided, and the computer readable storage medium is used to store a program, and the program is used to execute the real-time fusion type millimeter wave imaging frame rate up conversion method mentioned above.

[0053] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, apparatus, or computer program product. Accordingly, embodiments of the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer-readable program code.

[0054] The foregoing detailed description of the application has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the application to the precise form disclosed, and obviously many modifications and variations are possible in light of the above teaching. It is intended that the scope of the application be governed by the claims and their equivalents.

Claims

1. An image dataset expansion method for a terahertz imaging security system, comprising the following steps: S100, a terahertz imaging security device collects an original terahertz imaging image of a target person; S200, each frame of the collected original terahertz imaging image is subjected to image expansion processing to obtain an expanded new image corresponding to each frame of the original terahertz imaging image; S300, each frame of the expanded new image is inserted into the corresponding original terahertz imaging image; In step S200, each frame of the original terahertz imaging image is subjected to image expansion processing, which comprises the following steps in turn: S210, the imaging contour of the target person in the current frame of the original terahertz imaging image is preprocessed, and the preprocessed image is taken as a to-be-processed image; S220, the geometric position information of the imaging contour of the target person in the current to-be-processed image is determined; S230, the offset direction of the target person in the expanded new image is predicted according to the geometric position information obtained in step S220; S240, the contour of the target person in the to-be-processed image is subjected to overall offset processing according to the offset direction predicted in step S230, to obtain a bias image based on the current single frame of the original terahertz imaging image; S250, the intersection of the bias image obtained in step S240 and the current frame of the original terahertz imaging image is processed, and then scaling processing is performed, to obtain the expanded new image based on the current single frame of the original terahertz imaging image.

2. The method of claim 1, wherein: In step S210, the imaging contour of the target person in the current frame of the original terahertz imaging image is preprocessed by applying a semantic segmentation technique to obtain the complete outer contour of the imaging of the target person in the current frame of the original terahertz imaging image.

3. The method of claim 2, wherein: In step S220, the circumscribed rectangle of the outer contour edge of the human body in the to-be-processed image obtained in step S210 is obtained, and the center point P position of the circumscribed rectangle is determined, the image center point Q position of the to-be-processed image is obtained, and the geometric position of the imaging contour of the target person in the current to-be-processed image relative to the image center point Q of the to-be-processed image is determined according to the positions of the center point P of the circumscribed rectangle and the image center point Q of the to-be-processed image.

4. The method of claim 3, wherein: A rectangular coordinate system is established on the to-be-processed image with the image center point Q as the origin, and the position of the center point P of the circumscribed rectangle in the rectangular coordinate system is determined.

5. The method of claim 4, wherein: The rectangular coordinate system divides the to-be-processed image into four regions, namely the upper left region, the upper right region, the lower left region, and the lower right region. In step S230, if the center point P of the circumscribed rectangle is located in the upper left region, it is predicted that the imaging of the target person in the expanded new image moves left in the horizontal direction and moves up in the vertical direction of the rectangular coordinate system; if the center point P of the circumscribed rectangle is located in the upper right region, it is predicted that the imaging of the target person in the expanded new image moves right in the horizontal direction and moves up in the vertical direction of the rectangular coordinate system; if the center point P of the circumscribed rectangle is located in the lower left region, it is predicted that the imaging of the target person in the expanded new image moves left in the horizontal direction and moves down in the vertical direction of the rectangular coordinate system; if the center point P of the circumscribed rectangle is located in the lower right region, it is predicted that the imaging of the target person in the expanded new image moves right in the horizontal direction and moves down in the vertical direction of the rectangular coordinate system.

6. The method of claim 1, wherein: Step S240, the whole shift processing of the measured target personnel contour in the image to be processed is to shift the portrait by 1-2 pixel distance.

7. The method of claim 1, wherein: The intersection processing in step S250, first, the offset image is superimposed and compared with the corresponding current single-frame original terahertz imaging data graph, and the non-overlapping part in the single-frame original terahertz imaging data graph is cut off, and then the remaining part image is zoomed to the size of the single-frame original terahertz imaging data graph to form the extended new image.

8. A computer apparatus, characterized by: It comprises a processor, a memory and a program; the program is stored in the memory, and the processor calls the program stored in the memory to execute the image data set expansion method for the terahertz imaging security inspection system in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that: A computer readable storage medium is used to store a program for executing the image data set expansion method for the terahertz imaging security inspection system in any one of claims 1-7.

Images