CN112435252B - A detection method for warhead fragment perforation and pit - Google Patents

Abstract

The invention provides a method for detecting breaking perforation and pit of a warhead, which belongs to the technical field of image processing and comprises the following steps: for an equivalent target plate image acquired after static explosion of a warhead, carrying out geometric correction and actual physical size recovery based on a homography matrix of photographic transformation; linearly dividing the equivalent target plate based on the vertex and the boundary; smoothing the equivalent target plate image based on the guided filtering; combining the change of the illumination of the flash lamp at different positions on the equivalent target plate, and generating a hyperboloid threshold value by adopting image gradients; dividing a broken piece perforation area with bright gray and a broken piece pit area with dark gray on an equivalent target plate by using a hyperboloid threshold value; the broken perforation and pit area are used as seed points with gray level similarity to perform area growth; and removing the pseudo target area by using the fragment perforation and pit area and shape characteristic criterion. The calculation method can save the state information of the broken perforation and the pit, accurately divide the broken perforation and the pit area, and conveniently and rapidly realize the detection of the broken perforation and the pit.

Description

A method for detecting warhead fragment perforations and pits

Technical Field

The invention belongs to the field of image processing technology, relates to the field of image processing technology of warhead fragments hitting equivalent target plates, and specifically relates to a warhead fragment perforation and pit detection method.

Background Art

Obtaining accurate fragmentation field parameters under static explosion conditions is the basis for constructing the power situation of the warhead and the prerequisite for analyzing the dynamic explosion state of the fragmentation field. Before the static explosion, equivalent target plates are arranged at a certain distance and angle around the warhead. Due to the uncertainty of the fragment dispersion characteristics, the number, position, perforation and pit area of the fragments distributed on each equivalent target plate after the explosion are completely different. It is necessary to quickly obtain and count the situation of each equivalent target plate being hit by fragments, such as the number of fragment perforations on the equivalent target plate, which is the premise and basis for calculating the fragment dispersion characteristics, and the fragment dispersion characteristics are an important indicator for assessing the power parameters of ammunition. For the target plate method of measuring fragmentation field parameters, it is urgent to propose a fast, all-round, and automatic statistical fragmentation data detection technology.

At present, the detection of fragment perforation on equivalent target plates is mainly carried out through manual detection. However, in general, the explosion radius of the warhead is large, the number of equivalent target plates is large, and the number of fragment perforations is also large, so the workload of fragment perforation detection is large. At the same time, when the fragment perforation situation is complicated, there is no unified judgment standard, and multiple people are required to confirm on site. Moreover, the status information of fragment perforation is not saved, and the specific location of the fragment perforation cannot be reproduced, which is not conducive to subsequent verification, thus affecting the objectivity, accuracy and reliability of the detection. Therefore, it is necessary to study a fragment perforation and pit detection method based on image processing technology.

In order to further objectively, accurately and reliably detect fragment perforations and pits, a warhead fragment perforation and pit detection method was invented. This calculation method can save the fragment perforation and pit state information, accurately segment the fragment perforation and pit area, and conveniently and quickly realize the detection of fragment perforations and pits.

Summary of the invention

In order to overcome the above-mentioned deficiencies in the prior art, the present invention provides a warhead fragment perforation and pit detection method.

In order to achieve the above object, the present invention provides the following technical solutions:

The fragment perforation and pit detection method based on hyperbolic threshold segmentation includes the following steps:

Step 1: Collect the first equivalent target plate image after the static explosion of the warhead, accurately select the four vertices of the first equivalent target plate image by manual interaction, and obtain the vertex image coordinates (x′ _i , y′ _i , 1) ^T ; combine the prior shape and physical size of the equivalent target plate to obtain the physical coordinates of the equivalent target plate vertices (x _i , y _i , 1) ^T , i = 1, 4, and establish and calculate the homography matrix H _3×3 of the photographic transformation:

Based on the homography matrix H _3×3, the first equivalent target plate image is geometrically corrected and the actual physical size is restored to obtain a second equivalent target plate image;

Step 2: based on the four vertices of the equivalent target plate and the boundary linearity in the second equivalent target plate image, segment the second equivalent target plate image to obtain a third equivalent target plate image p _i ;

Step 3: based on guided filtering, smooth the third equivalent target image p _i to obtain a fourth equivalent target image q _i , and perform guided filtering on the fourth equivalent target image q _i ;

The fourth equivalent target plate image after guided filtering is:

q _i ′＝a _k I _i +bi∈ω _k (2)

Where, _qi ′ is the output fourth equivalent target image, _Ii is the guide image, here the third equivalent target image _pi is used as the guide image, _ak and b are the invariant coefficients of the linear function when the window center is located at k, _ak and b are the cost function E that minimizes the difference between images _pi and _qi ′:

Step 4: combining the changes of the flash light on different positions of the equivalent target plate arranged on site, using the gradient of the fourth equivalent target plate image q _i ′ after guided filtering to generate a hyperbolic threshold;

Step 5: using a hyperbolic threshold to segment the fourth equivalent target plate image q _i ′ after the filtering to obtain a bright gray fragment perforation area and a dark gray fragment pit area;

Step 6: Use the fragment perforation and pit areas as seed points of grayscale similarity to perform region growing;

Step 7: Use the fragment perforation and pit area and shape feature criteria to remove the false target area.

Preferably, the step 4 specifically includes:

Step 4.1, calculate the grayscale gradient of the fourth equivalent target plate image _qi ′ after guided filtering based on the canny operator; for the gradient that is positive first and then negative, take the absolute value, and for the gradient that is negative first and then positive, set it to 0, and use the mean value to calculate the first gradient threshold;

Step 4.2, using a thinning algorithm to obtain edge points, and using the grayscale value of the smoothed image corresponding to the edge points as the first reference threshold;

Step 4.3: For the gradient that is negative first and then positive, take the absolute value, and for the gradient that is positive first and then negative, set it to 0, and calculate the second gradient threshold based on the mean;

Step 4.4, using a thinning algorithm to obtain edge points; using the grayscale value of the smoothed image corresponding to the edge points as the second reference threshold;

Step 4.5: For the first equivalent target plate image taken with flash lighting, establish a world coordinate system OXYZ with the upper left corner of the space equivalent target plate as the origin and two vertical sides as coordinate axes; according to the camera imaging model, combined with the prior shape and size of the equivalent target plate, use the homography matrix to calculate the rotation matrix R = [r ₁ , r ₂ , r ₃ ] and the translation vector t of the camera coordinate system relative to the world coordinate system, and then according to the formula:

The three-dimensional coordinates of each pixel point on the first equivalent target plate image are converted from the world coordinate system [XYZ] ^T to the camera coordinate system [xyz] ^T , thereby obtaining the distance d from each pixel point on the first equivalent target plate image to the camera optical center (x ₀ , y ₀ , z ₀ ):

The size of the flash light source is relatively large and far away from the object being photographed, so it can be approximated as a point light source, and its position is approximately at the optical center of the camera. The relationship between the imaging grayscale f(x, y) of the flash light source illumination in the image and the distance d is:

Thus, the grayscale change ratio of different positions of the first equivalent target plate image relative to the center position of the imaging object is given;

Step 4.6: Based on the reference threshold 1 and the second reference threshold, combined with the grayscale change ratio of the flash illumination, generate a first curved surface threshold and a second curved surface threshold.

Preferably, the step 5 specifically includes:

Step 5.1, using the first curved surface threshold to segment the fourth equivalent target plate image q _i ′ after the guide filtering, the grayscale of the pixels greater than the double first curved surface threshold is set to 1, otherwise it is set to 0, to obtain the first segmented image of the bright grayscale fragment perforation area;

Step 5.2, using the double second curved surface threshold segmentation to guide the fourth equivalent target plate image q _i ′ after filtering, setting the grayscale of pixels less than the second curved surface threshold to 1, otherwise set to 0, to obtain the second segmentation image of the dark gray fragment pit area;

Step 5.3, performing an AND operation on the segmented image 1 and the second segmented image to obtain a third segmented image including the fragment perforation and fragment pit area.

Preferably, the step 6 specifically includes:

For the third segmented image, the fragment perforation and pit area is taken as the seed point area, and a single pixel expansion pre-operation is performed around it; if the difference between the expanded pixel and the grayscale mean of the fourth equivalent target plate image _qi ′ after the guided filtering corresponding to the seed point area is within the threshold criterion R, the expanded pixel is set to 1 and regional growth is performed, otherwise regional growth is not performed; the expansion pre-operation and regional growth are repeated until there is no pixel growth and finally a growth image is obtained.

Preferably, the step 7 specifically includes:

For the fragment perforation and pit regions in the growth image that contain pseudo targets, determine the range criterion _Wn for the number of pixels contained in the fragment perforations and pits, as well as the range criterion _Wb for the shape ratio of their major axis to minor axis. Calculate the number of pixels and shape ratio of each fragment perforation or pit region and compare them with the range criterion. If they are outside the criterion range, they are pseudo targets and are removed.

A warhead fragment perforation and pit detection method provided by the present invention has the following beneficial effects: for imaging noise with large amplitude and density caused by the roughness of an equivalent target plate, guided filtering is used to obtain a denoised image with protected edges; for situations where background grayscale is inconsistent and target attributes are different, the fragment perforation and pit areas are simultaneously and accurately segmented and detected in combination with a flash light illumination function, and the actual position coordinates of the fragment perforations and pits can be calculated by geometric correction of shooting the equivalent target plate in any posture; the present invention objectively, accurately and reliably detects fragment perforations and pits, and the calculation method can save fragment perforation and pit state information, accurately segment fragment perforation and pit areas, and conveniently and quickly realize the detection of fragment perforations and pits.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the embodiment of the present invention and its design scheme, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

FIG1 is a flow chart of a method for detecting warhead fragment perforations and dents according to Embodiment 1 of the present invention.

DETAILED DESCRIPTION

In order to enable those skilled in the art to better understand the technical solution of the present invention and implement it, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the scope of protection of the present invention.

Example 1

The present invention provides a warhead fragment perforation and pit detection method, as shown in FIG1, comprising the following steps:

Step 1: collect the first equivalent target plate image after the static explosion of the warhead, accurately select the four vertices of the first equivalent target plate image by manual interaction, perform geometric correction and actual physical size restoration on the homography matrix of the equivalent target plate based on photographic transformation in combination with the prior shape and size, and obtain the vertex image coordinates (x′ _i , y′ _i , 1) ^T ; obtain the equivalent target plate vertex physical coordinates (x _i , y _i , 1) ^T , i=1,...4, and establish and calculate the homography matrix H _3×3 of photographic transformation:

Based on the homography matrix H _3×3, the first equivalent target plate image is geometrically corrected and the actual physical size is restored to obtain a second equivalent target plate image;

Step 2: based on the four vertices of the equivalent target plate and the boundary linearity in the second equivalent target plate image, segment the second equivalent target plate image to obtain a third equivalent target plate image p _i ;

Step 3: based on guided filtering, smooth the third equivalent target image p _i to obtain a fourth equivalent target image q _i , and perform guided filtering on the fourth equivalent target image q _i ;

The fourth equivalent target plate image after guided filtering is:

q _i ′＝a _k I _i +bi∈ω _k (2)

Where, _qi ′ is the output fourth equivalent target image, _Ii is the guide image, here the third equivalent target image _pi is used as the guide image, _ak and b are the invariant coefficients of the linear function when the window center is located at k, _ak and b are the cost function E that minimizes the difference between images _pi and _qi ′:

Step 4: combining the changes of the flash light on different positions of the equivalent target plate arranged on site, using the gradient of the fourth equivalent target plate image q _i ′ after guided filtering to generate a hyperbolic threshold;

Specifically, in this embodiment, step 4 specifically includes:

Step 4.1, calculate the grayscale gradient of the fourth equivalent target plate image _qi ′ after guided filtering based on the canny operator; for the gradient that is positive first and then negative, take the absolute value, and for the gradient that is negative first and then positive, set it to 0, and use the mean value to calculate the first gradient threshold;

Step 4.2, using a thinning algorithm to obtain edge points, and using the grayscale value of the smoothed image corresponding to the edge points as the first reference threshold;

Step 4.3: For the gradient that is negative first and then positive, take the absolute value, and for the gradient that is positive first and then negative, set it to 0, and calculate the second gradient threshold based on the mean;

Step 4.4, using a thinning algorithm to obtain edge points; using the grayscale value of the smoothed image corresponding to the edge points as the second reference threshold;

Step 4.5: For the first equivalent target plate image taken with flash lighting, establish a world coordinate system OXYZ with the upper left corner of the space equivalent target plate as the origin and two vertical sides as coordinate axes; according to the camera imaging model, combined with the prior shape and size of the equivalent target plate, use the homography matrix to calculate the rotation matrix R = [r ₁ , r ₂ , r ₃ ] and the translation vector t of the camera coordinate system relative to the world coordinate system, and then according to the formula:

The three-dimensional coordinates of each pixel point on the first equivalent target plate image are converted from the world coordinate system [XYZ] ^T to the camera coordinate system [xyz] ^T , thereby obtaining the distance d from each pixel point on the first equivalent target plate image to the camera optical center (x ₀ , y ₀ , z ₀ ):

The size of the flash light source is relatively large and far away from the object being photographed, so it can be approximated as a point light source, and its position is approximately at the optical center of the camera. The relationship between the imaging grayscale f(x, y) of the flash light source illumination in the image and the distance d is:

Thus, the grayscale change ratio of different positions of the first equivalent target plate image relative to the center position of the imaging object is given;

Step 4.6: Based on the reference threshold 1 and the second reference threshold, combined with the grayscale change ratio of the flash illumination, generate a first curved surface threshold and a second curved surface threshold.

Step 5, using a hyperbolic threshold to segment the fourth equivalent target plate image _qi ′ after the guide filtering to obtain the bright gray fragment perforation area and the dark gray fragment pit area;

Specifically, in this embodiment, step 5 specifically includes:

Step 5.1, using the first curved surface threshold to segment the fourth equivalent target plate image q _i ′ after the guide filtering, the grayscale of the pixels greater than the double first curved surface threshold is set to 1, otherwise it is set to 0, to obtain the first segmented image of the bright grayscale fragment perforation area;

Step 5.2, using the double second curved surface threshold segmentation to guide the fourth equivalent target plate image q _i ′ after filtering, setting the grayscale of pixels less than the second curved surface threshold to 1, otherwise set to 0, to obtain the second segmentation image of the dark gray fragment pit area;

Step 5.3, performing an AND operation on the segmented image 1 and the second segmented image to obtain a third segmented image including the fragment perforation and fragment pit area.

Step 6: Use the fragment perforation and pit areas as seed points of grayscale similarity to perform region growing;

Specifically, in this embodiment, step 6 specifically includes:

For the third segmented image, the fragment perforation and pit area is taken as the seed point area, and a single pixel expansion pre-operation is performed around it; if the difference between the expanded pixel and the grayscale mean of the fourth equivalent target plate image q _i ′ after the guided filtering corresponding to the seed point area is within the threshold criterion R, the expanded pixel is set to 1 and regional growth is performed, otherwise regional growth is not performed; the expansion pre-operation and regional growth are repeated until there is no pixel growth and finally a growth image is obtained.

Step 7: Use the fragment perforation and pit area and shape feature criteria to remove the false target area.

Specifically, in this embodiment, step 7 specifically includes:

For the fragment perforation and pit regions in the growth image that contain pseudo targets, determine the range criterion _Wn for the number of pixels contained in the fragment perforations and pits, as well as the range criterion _Wb for the shape ratio of their major axis to minor axis. Calculate the number of pixels and shape ratio of each fragment perforation or pit region and compare them with the range criterion. If they are outside the criterion range, they are pseudo targets and are removed.

The detection method proposed in this embodiment objectively, accurately and reliably detects fragment perforations and pits, and uses the homography matrix of projection transformation to perform geometric correction on the equivalent target plate, thereby ensuring the consistency of the effect of shooting the equivalent target plate in any posture; the equivalent target plate is segmented based on the linearity of vertices and boundaries to remove complex background interference; based on guided filtering, denoising is performed while protecting the edge; combined with the changes in the flash light shining on different positions of the equivalent target plate, the image gradient is used to generate a hyperbolic threshold, while taking into account the bright gray fragment perforations and dark gray fragment pits; the hyperbolic threshold is used to segment the bright gray fragment perforation area and the dark gray fragment pit area on the equivalent target plate, respectively, to effectively extract seed points; the fragment perforation and pit area are used as seed points of grayscale similarity for regional growth, and the target area is accurately obtained; the fragment perforation and pit area and shape feature criteria are used to remove the pseudo target area, obtain the effective target area, accurately segment the fragment perforation and pit area, and conveniently and quickly realize the detection of fragment perforations and pits.

The embodiments described above are only preferred specific implementation modes of the present invention, and the protection scope of the present invention is not limited thereto. Any simple changes or equivalent replacements of the technical solutions that can be obviously obtained by any technician familiar with the field within the technical scope disclosed in the present invention belong to the protection scope of the present invention.

Claims (5)

1. The method for detecting the perforation and pit of the broken piece of the warhead is characterized by comprising the following steps:

step 1, acquiring a first equivalent target plate image after static explosion of a warhead, selecting four vertexes of the first equivalent target plate image in a manual interaction mode, and acquiring vertex image coordinates (x '' _i ,y′ _i ,1) ^T The method comprises the steps of carrying out a first treatment on the surface of the Combining the prior shape and the physical size of the equivalent target plate to obtain the equivalent target plate topPoint physical coordinates (x _i ,y _i ,1) ^T I=1,..4, building and calculating homography matrix H of photographic transformation _3×3 ：

Based on homography matrix H _3×3 Performing geometric correction and actual physical size recovery on the first equivalent target plate image to obtain a second equivalent target plate image;

step 2, dividing the second equivalent target plate image based on the linearity of four vertexes and boundaries of the equivalent target plate in the second equivalent target plate image to obtain a third equivalent target plate image p _i ；

Step 3, for the third equivalent target plate image p _i Smoothing to obtain a fourth equivalent target plate image q _i And for the fourth equivalent target plate image q _i Conducting guide filtering;

the fourth equivalent target plate image after the guide filtering is as follows:

q′ _i ＝a _k I _i +b (2)

wherein q _i ' fourth equivalent target plate image as output, I _i For guiding the image, i.e. the third equivalent target plate image p _i ，a _k And b is the invariant coefficient of the linear function when the window center is at k;

step 4, combining the change of different positions of the flash lamp illumination on the equivalent target plate arranged on site, and adopting a fourth equivalent target plate image q _i ' gradient generation hyperboloid threshold;

step 5, segmenting the fourth equivalent target plate image q by using hyperboloid threshold value _i The obtained bright gray fragment perforation area and dark gray fragment pit area;

step 6, taking the broken perforation and the pit area as seed points with gray level similarity for area growth;

and 7, removing the pseudo target area by using the fragment perforation and pit area and shape characteristic criterion.

2. The warhead fragment perforation and pit detection method of claim 1, wherein the step 4 specifically comprises:

step 4.1, calculating a fourth equivalent target plate image q based on a canny operator _i ' gray scale gradient; taking an absolute value for a gradient which is positive and negative firstly, setting 0 for the gradient which is negative and positive firstly, and calculating a first gradient threshold value by adopting an average value;

step 4.2, obtaining edge points of a first gradient threshold by adopting a thinning algorithm, and taking a gray value of a smooth image corresponding to the edge points of the first gradient threshold as a first reference threshold;

step 4.3, taking an absolute value for the gradient which is negative before positive, setting 0 for the gradient which is positive before negative, and calculating a second gradient threshold value based on the average value;

step 4.4, obtaining edge points of a second gradient threshold value by adopting a thinning algorithm; taking the gray value of the smooth image corresponding to the edge point of the second gradient threshold value as a second reference threshold value;

step 4.5, for a first equivalent target plate image shot by using flash lamp illumination, establishing a world coordinate system OXYZ by taking the upper left corner of a space equivalent target plate as an original point and two vertical sides as coordinate axes; according to the camera imaging model, combining the prior shape and the prior size of the equivalent target plate, adopting a homography matrix to calculate a rotation matrix R= [ R ] of a camera coordinate system relative to a world coordinate system ₁ ,r ₂ ,r ₃ ]And a translation vector t, then according to the equation:

three-dimensional coordinates of each pixel point on the first equivalent target plate image are obtained from a world coordinate system [ X Y Z] ^T Conversion to camera coordinate system [ x y z ]] ^T Thereby obtaining the image of each pixel point on the first equivalent target plate to the optical center (x ₀ ,y ₀ ,z ₀ ) Distance d of (2):

the relation between the imaging gray scale f (x, y) of the flash light source illumination in the image and the distance d is as follows:

thereby giving gray scale variation ratios of different positions of the first equivalent target plate image relative to the center position of the imaging object;

and 4.6, generating a first curved surface threshold value and a second curved surface threshold value based on the first reference threshold value and the second reference threshold value and combining the gray scale change proportion of the flash illumination.

3. The warhead fragment perforation and pit detection method of claim 2, wherein the step 5 specifically comprises:

step 5.1, segmenting the fourth equivalent target plate image q by utilizing the first curved surface threshold value _i Setting the pixel gray level larger than the threshold value of the double first curved surfaces as 1, otherwise setting the pixel gray level as 0, and obtaining a first segmentation image of the bright gray level fragment perforation area;

step 5.2, guiding the filtered fourth equivalent target plate image q by utilizing the threshold segmentation of the double second curved surfaces _i Setting the pixel gray level smaller than the threshold value of the second curved surface to be 1, otherwise setting the pixel gray level to be 0, and obtaining a second segmentation image of the dark gray level fragment pit area;

and 5.3, performing AND operation on the segmented image 1 and the second segmented image to obtain a third segmented image containing the broken piece perforation and the broken piece pit area.

4. The warhead fragment perforation and pit detection method of claim 3, wherein the step 6 specifically comprises:

for the third segmented image, taking the fragment perforation and pit area as seed point areas, and performing single pixel expansion pre-operation on the periphery of the fragment perforation and pit area; if the expansion pixel corresponds to the seed point regionFourth equivalent target plate image q after conducting filtering _i Setting the expanded pixel as 1 if the difference between the gray scale of the pixel and the gray average value of the gray scale is within the range of a threshold criterion R, and carrying out region growth, otherwise, not carrying out region growth; the dilation pre-operation and region growing are repeated until no pixel growth stops, and finally a growing image is obtained.

5. The warhead fragment perforation and pit detection method of claim 4, wherein the step 7 specifically comprises:

for the broken perforation and pit area in the growth image to contain pseudo target, determining the range criterion W of the pixel number contained in the broken perforation and pit _n And shape ratio range criterion W of their major axis and minor axis _b The number of pixels and the shape ratio of each fragment perforation or pit area are calculated and compared with a range criterion, and if the number of pixels and the shape ratio are out of the range criterion, the number of pixels and the shape ratio are pseudo targets and removed.

Images