CN115311685A - Millimeter wave image detection result judgment method based on average structure similarity - Google Patents
Millimeter wave image detection result judgment method based on average structure similarity Download PDFInfo
- Publication number
- CN115311685A CN115311685A CN202210938981.9A CN202210938981A CN115311685A CN 115311685 A CN115311685 A CN 115311685A CN 202210938981 A CN202210938981 A CN 202210938981A CN 115311685 A CN115311685 A CN 115311685A
- Authority
- CN
- China
- Prior art keywords
- detection result
- human body
- millimeter wave
- image
- contraband
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The invention discloses a millimeter wave image detection result judgment method based on average structure similarity, which comprises the steps of firstly, binary segmentation of a human body millimeter wave image to obtain a human body region binary image in the human body millimeter wave image; then calculating the position of the human body middle axis to obtain the horizontal coordinate of the human body middle axis; according to the obtained human body region binary image, performing specificity correction on a detection result; then carrying out mirror image processing on the detected result; and finally, searching whether a structure similar to the detection result exists on the other side of the human body with the central axis of the human body as the symmetry axis, and judging whether the detection result contains a contraband target according to the similarity degree. The method and the device effectively judge the target detection result of the millimeter wave image of the human body by using the SSIM algorithm in combination with the symmetrical property of the human body aiming at the problem that the detection result of the target detection of the millimeter wave image of the human body is difficult to judge whether the detection result really contains the target of the contraband because the human body area of the millimeter wave image of the human body is often covered with random noise which is difficult to remove.
Description
Technical Field
The invention belongs to the field of human body millimeter wave image target detection, and particularly relates to a millimeter wave image detection result judgment method based on average structure similarity.
Background
Millimeter wave image target detection is the key to realize the detection of contraband carried on the body surface of a human body, can be widely applied to security inspection work of airports, stations and the like, and is an effective substitute for the existing human body security inspection means. The imaging of the human body by using the millimeter waves is a premise for realizing millimeter wave image target detection, and the active millimeter wave imaging technology irradiates the millimeter waves to the human body and receives millimeter wave echoes by using a millimeter wave radar to generate an image according to the strength difference of the echoes.
Structural similarity (SSIM, zhou W, bovik A C, sheikh H R, et al. Image quality assessment: from error visibility to structural similarity [ J ]. IEEE Trans Image Process,2004,13 (4).) is commonly used as an index for assessing the similarity of two images in the field of Image processing, and the theory calculates assessment parameters by comparing the brightness, contrast and structure of two images, and the parameters are used for assessing the similarity of two images. Due to the fact that the millimeter wave imaging device is unstable and the like, random noise which is difficult to remove is covered in a human body area of a human body millimeter wave image, and therefore it is difficult to judge whether a detection result of target detection really contains a contraband target or not.
Aiming at the problem, the target detection result of the millimeter wave image of the human body is effectively judged by using an SSIM algorithm in combination with the symmetric property of the human body.
Disclosure of Invention
The invention provides a millimeter wave image detection result judgment method based on structural similarity, aiming at the problem that the detection result of target detection of a human body millimeter wave image is difficult to judge whether the detection result really contains a contraband target or not because the human body area of the human body millimeter wave image is often covered with random noise which is difficult to remove.
A millimeter wave image detection result judgment method based on average structure similarity comprises the following steps:
step 1, binary segmentation is carried out on the human body millimeter wave image to obtain a human body region binary image in the human body millimeter wave image.
And 2, calculating the position of the human body middle axis to obtain the horizontal coordinate of the human body middle axis.
Step 3, performing specificity correction on the detection result according to the human body region binary image obtained in the step 1;
step 4, carrying out mirror image processing on the detection result;
step 5, calculating the structural similarity SSIM between the mirror image detection result and a sliding window region by using a sliding window method in the horizontal direction of the symmetrical position of the detection result relative to the axis of the human body, and recording the maximum value of SSIM;
and searching whether a structure similar to the detection result exists on the other side of the human body with the central axis of the human body as the symmetry axis, and judging whether the detection result contains a contraband target according to the similarity degree.
Further, in the step 1, a maximum inter-class variance method is used to perform binary segmentation on the human body region in the human body millimeter wave image to obtain a human body region binary image.
Further, the specific method of step 2 is as follows;
and establishing a plane rectangular coordinate system by taking the upper left corner point of the image as an original point O, wherein the downward direction and the rightward direction of the original point are respectively the positive directions of an x axis and a y axis.
And respectively taking points at the outermost edges of the left shank and the right shank of the human body under the same ordinate, and respectively recording the abscissa of the left shank and the abscissa of the right shank as legLeft and legRight. Let the abscissa of the medial axis of the human body be axis, then axis is expressed as:
further, the specific method in step 3 is as follows;
the purpose of the specificity correction is to reduce as much as possible the black background area within the detection result, i.e., the area that must be a non-contraband target. Specificity can be expressed as the ratio of the area of the contraband target to the area of the detection result:
wherein sp i Indicates the specificity of the i-th detection result, S obj Representing the area of the contraband object, S out The area of the detection result is shown.
As can be seen from the formula (2), for a detected object containing contraband, sp is the specific correction for the detected object i Certain rise is carried out; for a detection result not containing contraband objects, sp i One is defined as 0.
Traversing four edges of the circumscribed rectangle of the detection result in the binary image of the human body region, if the edges do not contain non-0 pixel points, contracting the edges towards the centroid direction of the detection result until the edges contain non-0 pixel points.
Further, the specific method of step 4 is as follows;
the human body structure is in mirror symmetry relative to the axis two sides, and the detected result needs to be subjected to mirror image processing firstly. The set of pixel points in the detection result is recorded as I out Hereinafter with I out To represent the detection result; creating an image with the same size as the detection result and recording as I Mirror And the initial gray-scale value is 0.
The mirror image processing specifically operates as follows: traversing I by taking a row of pixel points as basic unit out Is shown by out The gray value of each column of pixel points is correspondingly stored to I according to the reverse direction of column traversal Mirror The other side in the horizontal direction. For example, if I out Starting from the leftmost first column of pixel points to traverse to the right, I should be set out Storing gray values of pixel points in the leftmost first row into I Mirror The rightmost first column of pixel points, and so on, and finally I out Traversing to the rightmost row of pixel points, and storing the gray value of the row of pixel points to I Mirror The first column of pixel points on the leftmost side.
Further, the specific method in step 5 is as follows;
based on the symmetry of the human body structure, if I out The target without contraband is present in the horizontal direction of the symmetrical side of the axis Mirror And finding similar areas by using a sliding window method.
Firstly, defining an initial window position for sliding window method, recording pixel point set in the initial window as I win ，I win Size and I of out And (5) the consistency is achieved. Note I out The minimum abscissa of (1) is xOut, the minimum ordinate is yOut, the horizontal width is wOut, and the vertical height is hOut, and if the minimum abscissa of the initial window is xWin, the minimum ordinate is yWin, the horizontal width is wWin, and the vertical height is hWin, then I win And I out The following relationships exist:
sliding the initial window to find I Mirror Possible similar regions of (a): in the horizontal direction, the pixel points with the step length of 1 and the maximum sliding distance of wWin are used, the initial window is respectively slid leftwards and rightwards at the position of the initial window, and the pixel point set in the window in the sliding process is recorded as
Calculating average structural similarity
Expressed as:
in the sliding process of the recording window
Has a maximum value of
From the formula (2), sp i The higher the detection result, the higher the detection result
The lower, i.e. the more likely the detection result area contains contraband targets; sp i 0, the result of the detection is
The higher, i.e., the detection result region may contain only the natural structure of the human body.
As can be seen from the above-described analysis,
and event "I out The objects containing contraband are in negative correlation, thereby defining the reliability of the detection result:
the reliability of the detection result Trust represents I out Probability of containing contraband object, trust ∈ [0,1 [ ]]. Setting Trust threshold as Trust threshold When Trust ∈ [0,Trust ∈ threshold ) When it is considered to be I out No contraband target exists inside; when Trust belongs to [ Trust ∈ ] threshold ,1]When it is considered to be I out Within which contraband targets are present.
The invention has the following beneficial effects:
aiming at the problem that the detection result of the target detection of the human body millimeter wave image is difficult to judge whether the detection result really contains the contraband target or not because the human body area of the human body millimeter wave image is often covered with random noise which is difficult to remove, the target detection result of the human body millimeter wave image is effectively judged by combining the symmetric property of the human body and using an SSIM algorithm.
Drawings
FIG. 1 is a schematic diagram of a millimeter wave binary image of a human body according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of an airspace coordinate system and key point locations according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a specific modification of an embodiment of the present invention;
FIG. 4 is a schematic diagram of a mirror process according to an embodiment of the invention;
FIG. 5 is a schematic view of a sliding window area and method according to an embodiment of the present invention;
Detailed Description
The method of the present invention is further illustrated with reference to the following figures and examples.
A millimeter wave image detection result judgment method based on average structure similarity specifically comprises the following steps:
step 1, binary segmentation of millimeter wave images of a human body;
since the background Gray value of the millimeter wave image of the human body is almost 0, the maximum inter-class variance Method (OTSU, otsu N.A Threshold Selection Method from Gray-Level Histograms [ J ]. IEEE Transactions on Systems Man & Cybernetics,2007,9 (1): 62-66.) can be used to perform binary segmentation on the human body region. As shown in fig. 1, fig. 1 (a) is a human body millimeter wave image original image, and fig. 1 (b) is a human body region binary image obtained by using the OTSU method for fig. 1 (a).
Step 2, calculating the position of the middle axis of the human body;
a plane rectangular coordinate system is established by taking the upper left corner point of the image as an original point O, the downward direction and the rightward direction of the original point are respectively the positive directions of an x axis and a y axis, and a schematic diagram of the coordinate system and the following key point positions are marked in figure 2.
And respectively taking points at the outermost edges of the left shank and the right shank of the human body under the same ordinate, and respectively recording the abscissa of the left shank and the abscissa of the right shank as legLeft and legRight. Let the abscissa of the medial axis of the human body be axis, then axis is expressed as:
step 3, referring to the human body region binary image obtained in the step 1, and performing specificity correction on a detection result;
the purpose of the specificity correction is to reduce as much as possible the black background area within the detection result, i.e., the area that must be a non-contraband target. Specificity can be expressed as the ratio of the area of the contraband target (number of pixel points) to the area of the detection result (number of pixel points):
wherein sp i Indicates the specificity of the i-th detection result, S obj Representing the area of the contraband object, S out The area of the detection result is shown.
From the formula (2), for a detected result containing contraband object, if its specificity is modified, sp i Rising to a certain extent; for a detection result not containing contraband objects, sp i One is set to 0.
Fig. 3 (a) is a partial image of the left ankle of the human body in fig. 1 (a), and the OTSU results are shown in fig. 3 (b). The dotted line frame in fig. 3 (b) represents a detection result, the specificity correction method is to traverse four sides of a circumscribed rectangle of the detection result in the binary image of the human body region, and if the side does not contain the non-0 pixel point, the side is shrunk towards the centroid direction of the detection result until the side contains the non-0 pixel point. The arrow direction in fig. 3 (b) indicates the contraction direction of the right and lower sides of the detection result, and the dotted frame in fig. 3 (c) indicates the specificity correction result of the detection result.
Step 4, mirror image processing is carried out on the detection result;
the human body structure is in mirror symmetry relative to the axis two sides, and the detected result needs to be subjected to mirror image processing firstly. The set of pixel points in the detection result is recorded as I out Hereinafter with I out To represent the detection result; creating an image with the same size as the detection result and recording as I Mirror And the initial gray-scale value is 0.
The mirror image processing specifically operates as follows: traversing I by using a row of pixel points as basic unit out Is shown by out The gray value of each column of pixel points is correspondingly stored to I according to the reverse direction of column traversal Mirror The other side in the horizontal direction. For example, if I out Starting from the leftmost first column of pixel points to traverse to the right, I should be set out The leftmost sideStoring a row of pixel gray values to I Mirror The rightmost first column of pixel points, and so on, and finally I out Traversing to the rightmost row of pixel points, and storing the gray value of the row of pixel points to I Mirror The first column of pixel points on the leftmost side. FIG. 4 is I out And I Mirror Example diagrams, I out And I Mirror The curved arrow line between shows I in the mirror image processing case out And I Mirror The corresponding relation of each column.
Step 5, calculating the structural similarity SSIM between the mirror image detection result and a sliding window region by using a sliding window method in the horizontal direction of the symmetrical position of the detection result relative to the axis of the human body, and recording the maximum value of SSIM;
SSIM for image x and image y is represented as follows:
wherein, mu x ,μ y Is the average pixel value, σ, of the image x, y x ,σ y Is the standard deviation, σ, of the image x, y xy Is the covariance of the image x, y, C 1 ,C 2 To prevent constants with denominators of 0.
In practical applications, SSIM is generally calculated for local blocks of an Image, and the average value of all blocks SSIM is used as the average structural similarity of the Image (MSSIM, zhou W, bovik a C, sheikh H R, et al. Image quality assessment: from Image visibility to structural similarity [ J ]. IEEE Trans Image processes, 2004,13 (4)), which is expressed as:
wherein MSSIM (x, y) represents the average structural similarity of the image x and the image y, M is the total block number of the image, x j Is the jth image block of image x. Average structure similarity MSSIM (x, y) epsilon [0,1 ] of image x and image y]The higher the MSSIM (x, y), the more similar the two images. Based on peopleSymmetry of the bulk structure, if I out Without contraband object (see fig. 4), there must be I in the horizontal direction of the symmetric side of axis Mirror And finding similar areas by using a sliding window method.
First, an initial window position is defined for the sliding window method, as shown in fig. 5, the solid-line frame area at the left ankle of the human body is a detection result I out And I with out The solid line frame symmetrical relative to axis is the initial window corresponding to the detection result, and the set of pixel points in the initial window is recorded as I win ，I win Size and I of out And (5) the consistency is achieved. Note I out The minimum abscissa of (1) is xOut, the minimum ordinate is yOut, the horizontal width is wOut, and the vertical height is hOut, and if the minimum abscissa of the initial window is xWin, the minimum ordinate is yWin, the horizontal width is wWin, and the vertical height is hWin, then I win And I out The following relationships exist:
sliding the initial window to find I Mirror Possible similar regions of (c): in the horizontal direction, the step length is 1 pixel point, the maximum sliding distance is wWin, and the initial window is respectively slid to the left and right at the initial window position (the dashed box area in FIG. 5 represents the sliding range of the initial window), and the pixel point set in the window in the sliding process is recorded as
Calculating average structural similarity
As can be seen from the formula (4),
expressed as:
in the sliding process of the recording window
Has a maximum value of
From the formula (2), sp i The higher the detection result, the higher the detection result
The lower, i.e. the more likely the detection result area contains contraband targets; sp i 0, the result of the detection is
The higher, i.e., the detection result region may contain only the natural structure of the human body.
As can be seen from the above-described analysis,
and event "I out The contraband object is contained in the system, and the negative correlation relationship is formed, so that the reliability of the detection result is defined as follows:
the reliability of the detection result Trust represents I out Probability of containing contraband object, trust ∈ [0,1 [ ]]. Setting Trust threshold as Trust threshold When Trust ∈ [0,Trust ∈ threshold ) When it is considered to be I out No contraband target exists inside; when Trust belongs to [ Trust ∈ ] threshold ,1]When it is considered to be I out Within which contraband targets are present.
Claims (6)
1. A millimeter wave image detection result judgment method based on average structure similarity is characterized by comprising the following steps:
step 1, binary segmentation is carried out on the human body millimeter wave image to obtain a human body region binary image in the human body millimeter wave image;
step 2, calculating the position of the human body middle shaft to obtain the horizontal coordinate of the human body middle shaft;
step 3, performing specificity correction on the detection result according to the human body region binary image obtained in the step 1;
step 4, carrying out mirror image processing on the detection result;
step 5, calculating the structural similarity SSIM between the mirror image detection result and a sliding window region by using a sliding window method in the horizontal direction of the symmetrical position of the detection result relative to the axis of the human body, and recording the maximum value of SSIM;
and searching whether a structure similar to the detection result exists on the other side of the human body with the middle axis of the human body as the symmetry axis, and judging whether the detection result contains a contraband target according to the similarity degree.
2. The method for determining the detection result of the millimeter wave image based on the average structural similarity as claimed in claim 1, wherein step 1 is to use the maximum inter-class variance method to perform binary segmentation on the human body region in the millimeter wave image of the human body to obtain a binary image of the human body region.
3. The method for determining the detection result of the millimetric wave image based on the average structural similarity as claimed in claim 2, wherein the step 2 is as follows;
establishing a plane rectangular coordinate system by taking the upper left corner point of the image as an original point O, wherein the downward direction and the rightward direction of the original point are respectively the positive directions of an x axis and a y axis;
respectively taking points at the outermost edges of the left shank and the right shank of the human body under the same ordinate, and respectively recording the abscissa of the left shank and the abscissa of the right shank as legLeft and legRight; let the abscissa of the medial axis of the human body be axis, then axis is expressed as:
4. the method for judging the detection result of the millimeter wave image based on the average structural similarity according to claim 3, wherein the specific method in step 3 is as follows;
the purpose of the specificity correction is to reduce the black background area in the detection result as much as possible, namely the area which is not a contraband target; specificity can be expressed as the ratio of the area of the contraband target to the area of the detection result:
wherein sp i Indicates the specificity of the ith detection, S obj Representing the area of the contraband object, S out Indicating the area of the detection result;
from the formula (2), for a detected result containing contraband object, if its specificity is modified, sp i Certain rise is carried out; for a detection result not containing contraband objects, sp i One is fixed to be 0;
traversing four edges of the circumscribed rectangle of the detection result in the binary image of the human body region, if the edges do not contain non-0 pixel points, contracting the edges towards the centroid direction of the detection result until the edges contain non-0 pixel points.
5. The method for judging the detection result of the millimeter wave image based on the average structural similarity according to claim 4, wherein the specific method in step 4 is as follows;
the human body structure is in a mirror symmetry relation relative to two sides of axis, and the detected result needs to be subjected to mirror image processing firstly; the set of pixel points in the detection result is recorded as I out Hereinafter with I out To represent the detection result; creating an image with the same size as the detection result and recording as I Mirror Setting the initial grey value to 0;
the mirror image processing specifically operates as follows: traversing I by using a row of pixel points as basic unit out A first reaction of out The gray value of each column of pixel points is traversed according to the reverse direction of the columnIs correspondingly stored to I Mirror The other side in the horizontal direction; for example, if I out Starting from the leftmost first column of pixel points to traverse to the right, I should be set out Storing gray values of pixel points in the leftmost first row into I Mirror The rightmost first column of pixel points, and so on, and finally I out Traversing to the rightmost row of pixel points, and storing the gray value of the row of pixel points to I Mirror The first column of pixel points on the leftmost side.
6. The method for determining the detection result of the millimeter wave image based on the average structural similarity according to claim 5, wherein the specific method in step 5 is as follows;
based on the symmetry of the human body structure, if I out The target without contraband is present in the horizontal direction of the symmetric side of the axis Mirror Searching a similar area by using a sliding window method;
firstly, defining an initial window position for sliding window method, recording pixel point set in the initial window as I win ，I win Size and I of out The consistency is achieved; note I out The minimum abscissa of (1) is xOut, the minimum ordinate is yOut, the horizontal width is wOut, and the vertical height is hOut, and if the minimum abscissa of the initial window is xWin, the minimum ordinate is yWin, the horizontal width is wWin, and the vertical height is hWin, then I win And I out The following relationships exist:
sliding the initial window to find I Mirror Possible similar regions of (a): in the horizontal direction, the step length is 1 pixel point, the maximum sliding distance is wWin, the initial window is respectively slid leftwards and rightwards at the initial window position, and the pixel point set in the window in the sliding process is recorded as
Calculating average structural similarity
Expressed as:
in the sliding process of the recording window
Has a maximum value of
From the formula (2), sp i The higher the detection result, the higher the detection result
The lower, i.e. the more likely the detection result area contains contraband targets; sp i 0, the result of the detection is
The higher the result, that is, the detection result region may only contain the natural structure of the human body;
as can be seen from the above-described analysis,
and event "I out The objects containing contraband are in negative correlation, thereby defining the reliability of the detection result:
the reliability of the detection result Trust represents I out Probability of containing contraband object, trust ∈ [0,1 [ ]](ii) a Setting Trust threshold as Trust threshold When Trust belongs to [0 threshold ) When the utility model is used, the water is discharged,is considered to be I out No contraband target exists; when Trust belongs to [ Trust ∈ [ ] threshold ,1]When it is considered to be I out Within which contraband targets are present.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210938981.9A CN115311685B (en) | 2022-08-05 | 2022-08-05 | Millimeter wave image detection result judging method based on average structural similarity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210938981.9A CN115311685B (en) | 2022-08-05 | 2022-08-05 | Millimeter wave image detection result judging method based on average structural similarity |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115311685A true CN115311685A (en) | 2022-11-08 |
| CN115311685B CN115311685B (en) | 2023-05-02 |
Family
ID=83861429
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210938981.9A Active CN115311685B (en) | 2022-08-05 | 2022-08-05 | Millimeter wave image detection result judging method based on average structural similarity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115311685B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050110672A1 (en) * | 2003-10-10 | 2005-05-26 | L-3 Communications Security And Detection Systems, Inc. | Mmw contraband screening system |
| US20080021502A1 (en) * | 2004-06-21 | 2008-01-24 | The Trustees Of Columbia University In The City Of New York | Systems and methods for automatic symmetry identification and for quantification of asymmetry for analytic, diagnostic and therapeutic purposes |
| CN105513035A (en) * | 2014-09-22 | 2016-04-20 | 北京计算机技术及应用研究所 | Method and system for detecting human body hidden item in passive millimeter wave image |
| WO2018035814A1 (en) * | 2016-08-25 | 2018-03-01 | 华讯方舟科技有限公司 | Millimetre wave image-based human body foreign matter detection method and system |
| JP2018060422A (en) * | 2016-10-06 | 2018-04-12 | 株式会社Ｓｏｋｅｎ | Object detection device |
| US20180181833A1 (en) * | 2014-08-25 | 2018-06-28 | Agency For Science, Technology And Research | Methods and systems for assessing retinal images, and obtaining information from retinal images |
| CN110334571A (en) * | 2019-04-03 | 2019-10-15 | 复旦大学 | A kind of millimeter-wave image human body method for secret protection based on convolutional neural networks |
| CN110533025A (en) * | 2019-07-15 | 2019-12-03 | 西安电子科技大学 | The millimeter wave human body image detection method of network is extracted based on candidate region |
| CN112819094A (en) * | 2021-02-25 | 2021-05-18 | 北京时代民芯科技有限公司 | Target detection and identification method based on structural similarity measurement |
| WO2022057914A1 (en) * | 2020-09-21 | 2022-03-24 | 同方威视技术股份有限公司 | Millimeter-wave human body security checking system and method based on double standing postures |
| CN114612655A (en) * | 2022-05-10 | 2022-06-10 | 北京圣点云信息技术有限公司 | Vein recognition algorithm transplanting method and device |
-
2022
- 2022-08-05 CN CN202210938981.9A patent/CN115311685B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050110672A1 (en) * | 2003-10-10 | 2005-05-26 | L-3 Communications Security And Detection Systems, Inc. | Mmw contraband screening system |
| US20080021502A1 (en) * | 2004-06-21 | 2008-01-24 | The Trustees Of Columbia University In The City Of New York | Systems and methods for automatic symmetry identification and for quantification of asymmetry for analytic, diagnostic and therapeutic purposes |
| US20180181833A1 (en) * | 2014-08-25 | 2018-06-28 | Agency For Science, Technology And Research | Methods and systems for assessing retinal images, and obtaining information from retinal images |
| CN105513035A (en) * | 2014-09-22 | 2016-04-20 | 北京计算机技术及应用研究所 | Method and system for detecting human body hidden item in passive millimeter wave image |
| WO2018035814A1 (en) * | 2016-08-25 | 2018-03-01 | 华讯方舟科技有限公司 | Millimetre wave image-based human body foreign matter detection method and system |
| JP2018060422A (en) * | 2016-10-06 | 2018-04-12 | 株式会社Ｓｏｋｅｎ | Object detection device |
| CN110334571A (en) * | 2019-04-03 | 2019-10-15 | 复旦大学 | A kind of millimeter-wave image human body method for secret protection based on convolutional neural networks |
| CN110533025A (en) * | 2019-07-15 | 2019-12-03 | 西安电子科技大学 | The millimeter wave human body image detection method of network is extracted based on candidate region |
| WO2022057914A1 (en) * | 2020-09-21 | 2022-03-24 | 同方威视技术股份有限公司 | Millimeter-wave human body security checking system and method based on double standing postures |
| CN112819094A (en) * | 2021-02-25 | 2021-05-18 | 北京时代民芯科技有限公司 | Target detection and identification method based on structural similarity measurement |
| CN114612655A (en) * | 2022-05-10 | 2022-06-10 | 北京圣点云信息技术有限公司 | Vein recognition algorithm transplanting method and device |
Non-Patent Citations (7)
| Title |
|---|
| HARISH BHASKAR等: "Posed Facial Expression Detection Using Reflection Symmetry and Structural Similarity" * |
| IBRAGIM R. ATADJANOV等: "Reflection Symmetry Detection via Appearance of Structure Descriptor" * |
| 刘冰清;杨风暴;刘英杰;: "区域选取与差异图融合的目标变化检测方法", 科学技术与工程 * |
| 张健;王卫民;唐洋;: "利用深度学习进行毫米波图像违禁物体识别" * |
| 张勇;金伟其;: "基于结构相似度与感兴趣区域的图像融合评价方法" * |
| 程秋菊;陈国平;王璐;管春;: "基于卷积神经网络的毫米波图像目标检测", 科学技术与工程 * |
| 詹维;马新星;徐子剑;: "基于超像素分割的红外盲元检测及校正", 红外技术 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115311685B (en) | 2023-05-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210350149A1 (en) | Lane detection method and apparatus,lane detection device,and movable platform | |
| US6658150B2 (en) | Image recognition system | |
| JP3307335B2 (en) | Vehicle region detection device and vehicle region verification method | |
| EP2085928B1 (en) | Detection of blobs in images | |
| Yoo et al. | Image matching using peak signal-to-noise ratio-based occlusion detection | |
| US7756295B2 (en) | Change region detection device and change region detecting method | |
| CN110717934B (en) | Anti-occlusion target tracking method based on STRCF | |
| US11669978B2 (en) | Method and device for estimating background motion of infrared image sequences and storage medium | |
| De Souza | Edge detection using sliding statistical tests | |
| CN107977973B (en) | Method and device for acquiring irradiation field boundary of beam limiter in medical diagnosis image | |
| CN110348307B (en) | Path edge identification method and system for crane metal structure climbing robot | |
| Czerwinski et al. | An approach to boundary detection in ultrasound imaging | |
| CN115345821A (en) | Steel coil binding belt loosening abnormity detection and quantification method based on active visual imaging | |
| CN113205494B (en) | Infrared small target detection method and system based on adaptive scale image block weighting difference measurement | |
| CN112435249B (en) | Dynamic small target detection method based on circumferential scanning infrared search system | |
| CN115311685A (en) | Millimeter wave image detection result judgment method based on average structure similarity | |
| CN112734666B (en) | SAR image speckle non-local mean suppression method based on similarity value | |
| CN115797773A (en) | Ultra-wideband radar target identification method based on edge detection and region segmentation | |
| CN115731221A (en) | Self-adaptive infrared small target detection method considering neighborhood anisotropy | |
| CN115376106A (en) | Vehicle type identification method, device, equipment and medium based on radar map | |
| CN112712476B (en) | Denoising method and device for TOF ranging and TOF camera | |
| Wang et al. | A novel segmentation algorithm for side-scan sonar imagery with multi-object | |
| CN107451574B (en) | Motion estimation method based on Haar-like visual feature perception | |
| Liu et al. | Lane line detection based on OpenCV | |
| CN116184407B (en) | Automatic marking method, device and equipment for SAR image radiation quality problem |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |