A method for mapping suspicious objects and puppets in millimeter wave images
Through artificial intelligence posture recognition algorithm and coordinate conversion technology, the problem of inaccurate mapping of suspicious items in millimeter-wave human security inspection equipment is solved, accurate automatic mapping in puppet images is realized, and the reliability and efficiency of security inspection is improved, reflecting the Caring for the inspected persons.
Patent Information
- Application Number
- CN202111520462.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-12-13
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2041-12-13
AI Technical Summary
The existing millimeter-wave human security equipment has inaccurate mapping of suspicious items on the doll, resulting in missed or repeated inspections.
The artificial intelligence pose recognition algorithm is used to extract the pose features of the millimeter-wave human body image and the puppet model, and through coordinate conversion technology, the location of suspicious objects is automatically mapped from the millimeter-wave image to the puppet image, ensuring the accuracy and efficiency of the mapping.
It realizes accurate and automatic mapping of the location of suspicious items, improves the reliability and efficiency of security inspections, protects the privacy of the inspected personnel, and reduces the discomfort of the inspected personnel.
Smart Images

Figure CN114612930B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data recognition, and in particular to a method for mapping a suspicious object doll in a millimeter wave image. Background Art
[0002] With the continuous advancement of the construction of the four types of airports, namely "safe airport, green airport, smart airport and humanistic airport" and the severe epidemic situation at home and abroad, millimeter wave body security inspection equipment adopts a contactless rapid security inspection mode, has the ability to detect metals and non-metals, is safe and harmless to the human body, and is highly consistent with the construction requirements of the four types of airports, namely safe, green, smart and humanistic, and is increasingly used in the field of airport security.
[0003] In order to protect the privacy of the inspected persons and show humanistic care for them, the security inspection devices currently on the market all use puppets to display the suspicious items carried by the inspected persons. The suspicious items are first marked in the millimeter wave image, and then the result is displayed in the puppet image. If the position mark is wrong, it will cause missed inspections or repeated inspections of the inspected persons. Therefore, the accuracy of the position mark of suspicious items in the puppet image is crucial to the reliability of the millimeter wave security inspection performance. To solve this problem, a method for mapping suspicious items to puppets in millimeter wave images is proposed under the premise of ensuring accurate and efficient mapping, which realizes automatic mapping of puppets at the positions of suspicious items. Summary of the invention
[0004] The present invention aims to solve the problem of inaccurate position mapping of suspicious items on a doll in millimeter wave human body security inspection equipment, and proposes a method for mapping suspicious items on a doll in a millimeter wave image. An artificial intelligence posture recognition algorithm is used to extract posture features of a millimeter wave human image and a doll model, and coordinate conversion technology is used to complete the automatic mapping of a millimeter wave suspicious item marking frame from a human image to a doll image. The method can accurately map the position of suspicious items from a millimeter wave image to a doll image, and effectively protect the personal privacy of the inspected person while facilitating security personnel to conduct manual inspections on the inspected person.
[0005] The present invention provides a method for mapping a suspicious object puppet in a millimeter wave image, comprising the following steps:
[0006] S1. Marking millimeter wave image posture key points and training models: Marking posture key points of existing millimeter wave security inspection images, extracting posture key point coordinate information, forming a schematic diagram of key parts of the security inspection image through connection relationships, and numbering posture key points at the same time, creating an artificial intelligence posture recognition model and training it to obtain a trained artificial intelligence posture recognition model;
[0007] S2, extracting the coordinate information of the key points of the puppet image posture: using the trained artificial intelligence posture recognition model to extract the coordinate information of the key points of the puppet image posture, forming a schematic diagram of the key parts of the puppet image through the connection relationship and numbering the key points of the posture according to the method of step S1;
[0008] S3, extracting the position information of the posture key points of the inspected millimeter wave security inspection image: when the inspected person is undergoing security inspection, the millimeter wave security inspection image is photographed, and the posture key points are marked and the coordinate information of the posture key points is extracted from the millimeter wave security inspection image using the trained artificial intelligence posture recognition model, and the posture key points are numbered according to the method of step S1 to determine whether the inspected person needs to adjust the posture. If not, proceed to step S4; if so, prompt the inspected person to adjust the posture and re-inspect;
[0009] S4, determining the association between the suspicious object and the attitude key points of the millimeter wave security inspection image; determining the reference part through the positional relationship between the rectangular frame of the suspicious object in the millimeter wave security inspection image and the attitude key points, and obtaining the proportional relationship rx, ry between the suspicious object and the attitude key points in the reference part;
[0010] S5. Map the suspicious object in the puppet image: determine the reference part and posture key points of the puppet image according to the number of the posture key points in the reference part, and map the suspicious object to the puppet image according to the proportional relationship rx, ry. The four vertices of the rectangular box of the suspicious object in the puppet image are [P′1, P′2, P′3, P′4], and the center point is M1[mx1, my1]. The mapping of the suspicious object puppet in the millimeter wave image is completed.
[0011] The method for mapping a suspicious object puppet in a millimeter wave image of the present invention is, as a preferred embodiment, in step S1, the order of numbering the posture key points is from bottom to top and from left to right.
[0012] The method for mapping a millimeter wave image to a suspicious object puppet of the present invention is, as a preferred embodiment, in step S1 and step S2, the posture key point numbers of the security inspection image correspond one-to-one with the posture key point numbers of the puppet image.
[0013] The method for mapping suspicious objects to figures in millimeter wave images of the present invention is, as a preferred embodiment, in step S3, the method for determining whether the person under inspection needs to adjust his posture is: calculating whether the length and relative position relationship of the human body parts meet the requirements, the human body parts including the forearm, upper arm, thigh and calf.
[0014] The method for mapping suspicious objects and figures in millimeter wave images of the present invention is preferably implemented in that step S4 comprises the following steps:
[0015] S41, the position of the suspicious object in the millimeter wave security inspection image is represented by a rectangular frame, where the upper left point of the rectangular frame is P1 (x1, y1), the upper right point is P2 (x2, y1), the lower left point is P3 (x1, y2), and the lower right point is P4 (x2, y2)];
[0016] S42, judging whether the rectangular frame intersects with the line connecting the two posture key points A (ax, ay) and B (bx, by) to determine whether the part where point A and point B are located is the reference part, traversing all posture key points, if the reference part cannot be determined, proceeding to step S43; if the reference part can be determined, proceeding to step S44;
[0017] S43, determine the closest posture key point and reference part by judging the distance between the center point M (mx, my) of the rectangular frame and the posture key points A (ax, ay) and B (bx, by), and the calculation method is as follows:
[0018]
[0019] Among them: AM , l BM , l AB are the lengths of the straight lines AM, BM, and AB, respectively; p is the semicircumference; and d is the distance from the center point to the part;
[0020] Traverse all posture key points until the reference part is determined;
[0021] S44, if A (ax, ay), B (bx, by) are the key points of the posture in the reference part, the calculated distances in the X and Y directions between the center point M and the starting point A of the reference part are the same as l AB The proportional relationship between rx and ry,
[0022] rx=(mx-ax) / l AB
[0023] ry=(my-ay) / l AB ,
[0024] Among them, l AB are the lengths of straight lines AB respectively.
[0025] The method for mapping a suspicious object puppet in a millimeter wave image of the present invention is, as a preferred embodiment, in step S42, the method for determining whether the rectangular frame intersects with the line connecting the two posture key points A (dx, ay) and B (bx, by) comprises:
[0026] If the intersection point xi1 of the line AB and the upper edge of the rectangular frame is: x1≤xi1≤x2, then they intersect, where
[0027] If the intersection point xi2 of the line AB and the lower edge of the rectangular frame is: x1≤xi2≤x2, then they intersect, where
[0028] If the intersection point yi1 of the line AB and the left line of the rectangular box is: y1≤yi2≤y2, then they intersect, where
[0029] If the intersection point yi2 of the line AB and the right line of the rectangular box is: y1≤yi2≤y2, then they intersect, where
[0030] The method for mapping suspicious objects and figures in millimeter wave images of the present invention is preferably implemented in step S5, where mx1=l A1B1 *rx+ax,
[0031] my1=l A1B1 *ry+ay;
[0032] Among them, A1 is the mapping point of the posture key point A (ax, ay) of the reference part in the puppet image, B1 is the mapping point of the posture key point B (bx, by) of the reference part in the puppet image, A1B1 is the length of the straight line A1B1.
[0033] like Figure 1 Shown is a method flow chart of the system, which is implemented through the following steps.
[0034] The first step is to mark the key points of the human body in the millimeter-wave security inspection image, extract the coordinate information of the key points of the human body, and carry out artificial intelligence posture recognition model training.
[0035] The second step is to extract the position information of the key points of the puppet image posture. The artificial intelligence posture recognition algorithm is used to extract the position information of the key nodes of the puppet image posture. The key parts of the human body image are formed through the connection relationship. The parts are numbered from bottom to top and from left to right, such as Figure 2 shown.
[0036] The third step is to extract the position information of the key points of the millimeter wave image posture. The artificial intelligence posture recognition algorithm is used for the millimeter wave image to extract the key points of the millimeter wave human body image posture. The first step method is used to number each part, such as Figure 3 By calculating whether the length and relative position of human body parts such as forearms, upper arms, thighs, and calves meet the relevant requirements of GB10000-1988 "Human Body Dimensions of Chinese Adults", if they meet, proceed to the next step. If not, the inspected person will be prompted to adjust his posture and re-check.
[0037] The fourth step is to determine the positional relationship between the suspicious object and the millimeter wave attitude key points.
[0038] 1) The location of the suspicious object is shown in FIG4 , represented by a rectangular box, and represented as an array consisting of four vertex coordinates [upper left point P1 (x1, y1), upper right point P2 (x2, y1), lower left point P3 (x1, y2), lower right point P4 (x2, y2)];
[0039] 2) Determine whether the rectangular frame intersects with the straight line AB formed by the key points of the posture A (ax, ay) and B (bx, by). The judgment principle is to calculate the intersection points of the upper edge line, lower edge line, left edge line, right edge line of the rectangular frame and the straight line respectively. If any of the intersection points is on the edge line, it is an intersection. The details are as follows:
[0040] a) Intersection with the upper edge If x1≤xi1≤x2, then they intersect;
[0041] b) Intersection with the lower line If x1≤xi2≤x2, then they intersect;
[0042] c) Intersection with the left line If y1≤yi2≤y2, then they intersect;
[0043] d) Intersection point with right line If y1≤yi2≤y2, then they intersect;
[0044] Traverse all the parts until the intersecting part is found as the reference part.
[0045] 3) If the intersecting part cannot be found through step 2), the nearest part is selected as the reference part by judging the distance between the center point M (mx, my) of the rectangular frame and the line connecting each part. The calculation method is as follows:
[0046]
[0047] Among them: AM , l BM , l AB The length of the straight lines AM, BM and AB respectively is p, which is the semi-circumference, and d is the distance from the center point to the part.
[0048] 4) Calculate the distance in the X and Y directions between the midpoint M of the suspicious object rectangle and the starting point A of the reference part and the length l of the part AB The proportional relationship between them.
[0049] rx=(mx-ax) / l AB
[0050] ry=(my-ay) / l AB
[0051] Step 5: Map the position of the suspicious object in the puppet image. Since the number of the human body parts extracted by the puppet image posture and the millimeter wave human body image posture at the same position is consistent, through the reference part number extracted in the third step and the proportional relationship rx, ry between the suspicious object and the reference part, the coordinate transformation can be performed between the starting point A1 and the end point B1 of the part with the same number in the puppet image to obtain the position of the center point M1 of the suspicious object frame in the puppet:
[0052] mx1=l A1B1 *rx+ax
[0053] my1=l A1B1 *ry+ay
[0054] Through the length and width of the original rectangle, the positions of the four vertices [P′1, P′2, P′3, P′4] of the rectangular frame in the puppet are calculated to complete the mapping of the millimeter wave suspicious object to the puppet, as shown in FIG4 .
[0055] This method can automatically map suspicious objects in the front and back images of millimeter-wave images to images of puppets within 0.1 seconds, effectively improving the reliability of security inspections and reflecting humanistic care for the inspected persons.
[0056] The present invention has the following advantages:
[0057] (1) This method uses an artificial intelligence posture recognition algorithm to extract the posture features of the millimeter wave human image and the puppet model, and uses coordinate conversion technology to complete the automatic mapping of the millimeter wave suspicious object marking frame from the human image to the puppet image. It has high accuracy and fast speed, effectively improving the reliability of millimeter wave security inspection, reducing the discomfort of the inspected personnel, and increasing the security inspection efficiency.
[0058] (2) For millimeter wave images of inspected persons in different standing postures, the position of the rectangular frame of the suspicious object can be accurately mapped to the puppet image in the same standing posture, which has strong adaptability. BRIEF DESCRIPTION OF THE DRAWINGS
[0059] Figure 1 A flow chart of a method for mapping suspicious objects and puppets in millimeter wave images;
[0060] Figure 2 A method for mapping a suspicious object puppet in a millimeter wave image is a schematic diagram of key parts of a puppet image;
[0061] Figure 3 A method for mapping suspicious objects and puppets in millimeter wave images is a schematic diagram of the key parts of millimeter wave images;
[0062] Figure 4a A method for mapping suspicious objects in millimeter wave images to human figures is shown as a schematic diagram of suspicious objects in millimeter wave images;
[0063] Figure 4b The present invention is a schematic diagram of a method for mapping suspicious objects in millimeter wave images into puppets, and mapping suspicious objects in millimeter wave images into puppets. DETAILED DESCRIPTION
[0064] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.
[0065] Example 1
[0066] like Figure 1 As shown, a method for mapping suspicious objects and puppets in millimeter wave images includes the following steps:
[0067] S1. Marking millimeter wave image posture key points and training models: marking posture key points of existing millimeter wave security inspection images, extracting posture key point coordinate information, forming a schematic diagram of key parts of security inspection images through connection relationships, and numbering posture key points at the same time, creating an artificial intelligence posture recognition model and training it to obtain a trained artificial intelligence posture recognition model; the order of posture key point numbering is from bottom to top and from left to right;
[0068] S2. Extracting coordinate information of key points of the puppet image posture: Using the trained artificial intelligence posture recognition model to extract coordinate information of key points of the puppet image posture, such as Figure 2 As shown, a schematic diagram of key parts of the puppet image is formed through connection relationships and the posture key points are numbered according to the method of step S1; in step S1 and step S2, the posture key point numbers of the security inspection image correspond to the posture key point numbers of the puppet image one by one;
[0069] S3. Extracting the position information of the key points of the posture of the inspected millimeter wave security inspection image: When the inspected person is undergoing security inspection, the millimeter wave security inspection image is taken, and the trained artificial intelligence posture recognition model is used to mark the key points of the posture from the millimeter wave security inspection image and extract the coordinate information of the key points of the posture, such as Figure 3 As shown, the numbering is performed according to the method of step S1 to determine whether the inspected person needs to adjust his posture. If not, proceed to step S4; if yes, prompt the inspected person to adjust his posture and re-check;
[0070] The method to determine whether the person being tested needs to adjust his posture is to calculate whether the length and relative position of the body parts meet the requirements. The body parts include the forearm, upper arm, thigh and calf;
[0071] S4, determining the association between the suspicious object and the key points of the millimeter wave security inspection image posture; Figure 4a As shown, the reference part is determined by the positional relationship between the rectangular frame of the suspicious object and the posture key points in the millimeter wave security inspection image, and the proportional relationship rx, ry between the suspicious object and the posture key points in the reference part is obtained;
[0072] S41, the position of the suspicious object in the millimeter wave security inspection image is represented by a rectangular frame, where the upper left point of the rectangular frame is P1 (x1, y1), the upper right point is P2 (x2, y1), the lower left point is P3 (x1, y2), and the lower right point is P4 (x2, y2)];
[0073] S42, judging whether the rectangular frame intersects with the line connecting the two posture key points A (ax, ay) and B (bx, by) to determine whether the part where point A and point B are located is the reference part, traversing all posture key points, if the reference part cannot be determined, proceeding to step S43; if the reference part can be determined, proceeding to step S44;
[0074] The method for determining whether the rectangular frame intersects with the line connecting the two posture key points A (ax, ay) and B (bx, by) includes:
[0075] If the intersection point xi1 of the line AB and the upper edge of the rectangular frame is: x1≤xi1≤x2, then they intersect, where
[0076] If the intersection point xi2 of the line AB and the lower edge of the rectangular frame is: x1≤xi2≤x2, then they intersect, where
[0077] If the intersection point yi1 of the line AB and the left line of the rectangular box is: y1≤yi2≤y2, then they intersect, where
[0078] If the intersection point yi2 of the line AB and the right line of the rectangular box is: y1≤yi2≤y2, then they intersect, where
[0079] S43, determine the closest posture key point and reference part by judging the distance between the center point M (mx, my) of the rectangular frame and the posture key points A (ax, ay) and B (bx, by), and the calculation method is as follows:
[0080]
[0081] Among them: AM , l BM , l AB are the lengths of the straight lines AM, BM, and AB, respectively; p is the semicircumference; and d is the distance from the center point to the part;
[0082] Traverse all posture key points until the reference part is determined;
[0083] S44, if A (ax, ay), B (bx, by) are the key points of the posture in the reference part, the calculated distances in the X and Y directions between the center point M and the starting point A of the reference part are the same as l AB The proportional relationship between rx and ry,
[0084] rx=(mx-ax) / l AB
[0085] ry=(my-ay) / l AB ,
[0086] Among them, l AB are the lengths of straight lines AB respectively;
[0087] S5. Map the suspicious object in the puppet image: Figure 4b As shown, the reference part and posture key points of the puppet image are determined according to the number of the posture key points in the reference part, and the suspicious object is mapped to the puppet image according to the ratio relationship rx, ry. The four vertices of the rectangular box of the suspicious object in the puppet image are [P′1, P′2, P′3, P′4], and the center point is M1[mx1, my1]. The mapping of the suspicious object puppet in the millimeter wave image is completed;
[0088] mx1=l A1B1 *rx+ax,
[0089] my1=l A1B1 *ry+ay;
[0090] Among them, A1 is the mapping point of the posture key point A (ax, ay) of the reference part in the puppet image, B1 is the mapping point of the posture key point B (bx, by) of the reference part in the puppet image, A1B1 is the length of the straight line A1B1.
[0091] Example 2
[0092] like Figure 1 As shown, a method for mapping suspicious objects and puppets in millimeter wave images is implemented through the following steps.
[0093] The first step is to mark the key points of the human body in the millimeter-wave security inspection image, extract the coordinate information of the key points of the human body, and carry out artificial intelligence posture recognition model training.
[0094] The second step is to extract the position information of the key points of the puppet image posture. The artificial intelligence posture recognition algorithm is used to extract the position information of the key nodes of the puppet image posture. The key parts of the human body image are formed through the connection relationship. The parts are numbered from bottom to top and from left to right, such as Figure 2 shown.
[0095] The third step is to extract the position information of the key points of the millimeter wave image posture. The artificial intelligence posture recognition algorithm is used for the millimeter wave image to extract the key points of the millimeter wave human body image posture. The first step method is used to number each part, such as Figure 3 By calculating whether the length and relative position of human body parts such as forearms, upper arms, thighs, and calves meet the relevant requirements of GB10000-1988 "Human Body Dimensions of Chinese Adults", if they meet, proceed to the next step. If not, the inspected person will be prompted to adjust his posture and re-check.
[0096] The fourth step is to determine the positional relationship between the suspicious object and the millimeter wave attitude key points.
[0097] (1) The location of the suspicious item is as follows: Figure 4a As shown, it is represented by a rectangular box and represented as an array consisting of four vertex coordinates [upper left point P1 (x1, y1), upper right point P2 (x2, y1), lower left point P3 (x1, y2), lower right point P4 (x2, y2)];
[0098] (2) Determine whether the rectangular frame intersects with the straight line AB formed by the key points of the posture A (ax, ay) and B (bx, by). The judgment principle is to calculate the intersection points of the upper edge line, lower edge line, left edge line, and right edge line of the rectangular frame with the straight line respectively. If any of the intersection points is on the edge line, it is an intersection. The details are as follows:
[0099] e) Intersection with the upper edge If x1≤xi1≤x2, then they intersect;
[0100] f) Intersection with the lower line If x1≤xi2≤x2, then they intersect;
[0101] g) Intersection with the left line If y1≤yi2≤y2, then they intersect;
[0102] h) Intersection point with right line If y1≤yi2≤y2, then they intersect;
[0103] Traverse all the parts until the intersecting part is found as the reference part.
[0104] (3) If the intersecting part cannot be found through step (2), the nearest part is selected as the reference part by judging the distance between the center point M (mx, my) of the rectangular frame and the line connecting each part. The calculation method is as follows:
[0105]
[0106] Where: l AM , l BM , l AB The length of the straight lines AM, BM and AB respectively is p, which is the semi-circumference, and d is the distance from the center point to the part.
[0107] (4) Calculate the distance in the X and Y directions between the midpoint M of the suspicious object rectangle and the starting point A of the reference part, which is the same as the length l of the part. AB The proportional relationship between them.
[0108] rx=(mx-ax) / l AB
[0109] ry=(my-ay) / l AB
[0110] Step 5: Map the position of the suspicious object in the puppet image. Since the number of the human body parts extracted by the puppet image posture and the millimeter wave human body image posture at the same position is consistent, through the reference part number extracted in the third step and the proportional relationship rx, ry between the suspicious object and the reference part, the coordinate transformation can be performed between the starting point A1 and the end point B1 of the part with the same number in the puppet image to obtain the position of the center point M1 of the suspicious object frame in the puppet:
[0111] mx1=l A1B1 *rx+ax
[0112] my1=l A1B1 *ry+ay
[0113] Through the length and width of the original rectangle, the positions of the four vertices [P′1, P′2, P′3, P′4] of the rectangle in the puppet are calculated to complete the mapping of the millimeter wave suspicious object to the puppet, as shown in Figure 4b shown.
[0114] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A method for mapping suspicious objects and puppets in millimeter wave images. Features: The following steps are involved: S1. Marking millimeter wave image posture key points and training model: marking posture key points of existing millimeter wave security inspection images, extracting posture key point coordinate information, forming a schematic diagram of key parts of the security inspection image through connection relationships, and numbering the posture key points, creating an artificial intelligence posture recognition model and training it to obtain a trained artificial intelligence posture recognition model; S2, extracting the coordinate information of the key points of the puppet image posture: using the trained artificial intelligence posture recognition model to extract the coordinate information of the key points of the puppet image posture, forming a schematic diagram of the key parts of the puppet image through connection relationships and numbering the key points of the posture according to the method of step S1; S3, extracting the position information of the key points of the posture of the inspected millimeter wave security inspection image: when the inspected person is undergoing security inspection, the millimeter wave security inspection image is photographed, and the posture key points are marked and the coordinate information of the posture key points is extracted from the millimeter wave security inspection image using the trained artificial intelligence posture recognition model, and the posture key points are numbered according to the method of step S1, and it is determined whether the inspected person needs to adjust the posture. If not, proceed to step S4; if so, prompt the inspected person to adjust the posture and re-inspect; S4, determining the association between the suspicious object and the posture key points of the millimeter wave security inspection image; determining the reference part through the positional relationship between the rectangular frame of the suspicious object in the millimeter wave security inspection image and the posture key points, and obtaining the proportional relationship rx, ry between the suspicious object and the posture key points in the reference part; S5. Map the suspicious object in the puppet image: determine the reference part and posture key points of the puppet image according to the numbering of the posture key points in the reference part, and map the suspicious object to the puppet image according to the proportional relationship rx, ry. The four vertices of the rectangular box of the suspicious object in the puppet image are [P'1, P'2, P'3, P'4], and the center point is M1 [mx1, my1]. The mapping of the suspicious object puppet in the millimeter wave image is completed.
2. A method for mapping suspicious objects and figures in millimeter wave images according to claim 1, Features: In step S1, the order of numbering the posture key points is from bottom to top and from left to right.
3. A method for mapping suspicious objects and figures in millimeter wave images according to claim 1, Features: In step S1 and step S2, the posture key point numbers of the security inspection image correspond one-to-one with the posture key point numbers of the puppet image.
4. A method for mapping suspicious objects and figures in millimeter wave images according to claim 1, Features: In step S3, the method for determining whether the person under inspection needs to adjust his posture is: calculating whether the length and relative position relationship of the human body parts meet the requirements, wherein the human body parts include the forearm, upper arm, thigh and calf.
5. A method for mapping suspicious objects and figures in millimeter wave images according to claim 1, Features: Step S4 includes the following steps: S41, representing the position of the suspicious object in the millimeter wave security inspection image by a rectangular frame, wherein the upper left point of the rectangular frame is P1 (x1, y1), the upper right point is P2 (x2, y1), the lower left point is P3 (x1, y2), and the lower right point is P4 (x2, y2)]; S42, judging whether the rectangular frame intersects with the line connecting the two posture key points A (ax, ay) and B (bx, by) to determine whether the part where point A and point B are located is the reference part, traversing all the posture key points, if the reference part cannot be determined, proceeding to step S43; if the reference part can be determined, proceeding to step S44; S43, by judging the distance between the center point M (mx, my) of the rectangular frame and the posture key points A (ax, ay), B (bx, by), the closest posture key point and the reference part are determined, and the calculation method is as follows: p=(l AM +l BM +l AB ) / 2 Among them: AM , l BM , l AB are the lengths of the straight lines AM, BM, and AB, respectively; p is the semicircumference; and d is the distance from the center point to the part; Traversing all the posture key points until the reference part is determined; S44, if A(ax, ay) and B(bx, by) are the posture key points in the reference part, calculate the distances between the center point M and the starting point A of the reference part in the X direction and the Y direction. AB The proportional relationship between rx and ry, rx=mx-ax) / l AB ry=my-ay) / l AB , Among them, l AB is the length of line AB.
6. A method for mapping suspicious objects and figures in millimeter wave images according to claim 5, Features: In step S42, the method for determining whether the rectangular frame intersects with the line connecting the two posture key points A (ax, ay) and B (bx, by) includes: If the intersection point xi1 of the line AB and the upper edge of the rectangular frame is: x1≤xi1≤x2, then they intersect, where: If the intersection point xi2 of the line AB and the lower edge of the rectangular frame is: x1≤xi2≤x2, then they intersect, where: If the intersection point yi1 of the line AB and the left line of the rectangular frame is: y1≤yi2≤y2, then they intersect, where If the intersection point yi2 of the line AB and the right line of the rectangular frame is: y1≤yi2≤y2, then they intersect, where 7. A method for mapping suspicious objects and figures in millimeter wave images according to claim 5, Features: In step S5, mx1=1 A1B1 *rx+ax, my1=l A1B1 *ry+ay; Wherein, A1 is the mapping point of the posture key point A (ax, ay) of the reference part in the puppet image, B1 is the mapping point of the posture key point B (bx, by) of the reference part in the puppet image, A1B1 is the length of the straight line A1B1.
