CN117132622A - Non-imaging real-time detection tracking method and system for fast moving object - Google Patents
Non-imaging real-time detection tracking method and system for fast moving object Download PDFInfo
- Publication number
- CN117132622A CN117132622A CN202311164191.0A CN202311164191A CN117132622A CN 117132622 A CN117132622 A CN 117132622A CN 202311164191 A CN202311164191 A CN 202311164191A CN 117132622 A CN117132622 A CN 117132622A
- Authority
- CN
- China
- Prior art keywords
- scene
- moving object
- modulation base
- sub
- axis direction
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000003384 imaging method Methods 0.000 title claims abstract description 26
- 238000011897 real-time detection Methods 0.000 title claims abstract description 24
- 239000011159 matrix material Substances 0.000 claims abstract description 17
- 238000004590 computer program Methods 0.000 claims description 6
- 230000004044 response Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 8
- 238000012545 processing Methods 0.000 abstract description 6
- 238000005070 sampling Methods 0.000 abstract description 3
- 238000001514 detection method Methods 0.000 description 10
- 238000005259 measurement Methods 0.000 description 8
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 239000012141 concentrate Substances 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000007634 remodeling Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012634 optical imaging Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/77—Determining position or orientation of objects or cameras using statistical methods
Abstract
The invention relates to the technical field of computer vision processing, and discloses a non-imaging real-time detection tracking method and a non-imaging real-time detection tracking system for a fast moving object, wherein the method comprises the following steps: constructing a Hadamard matrix according to the scene size, remolding to obtain a modulation base sequence, optimizing the modulation base sequence, decomposing and reconstructing to obtain a sub-modulation base sequence; reconstructing a projection curve without a moving object and a projection curve with a moving object in the X-axis or Y-axis direction through the projection value of the sub-modulation base and the measured value acquired by the single-pixel detector, and calculating the first-order gradient and the first-order gradient difference of the projection curve; and determining the position of the fast moving object in the scene according to the gradient difference result. Under the condition that the target scene image is not required to be acquired, the spatial light modulator and the single-pixel detector are utilized to acquire the key information of the target space position, so that the method has the advantages of ultralow sampling rate and higher calculation efficiency, and can realize real-time detection and tracking of a rapid target.
Description
Technical Field
The invention relates to the technical fields of computer vision, image processing and optical imaging, in particular to a Single-Pixel Detector (Single-Pixel Detector) -based rapid moving object non-imaging real-time detection tracking method and system.
Background
Radar is an image-free target detection and tracking technology, is commonly used for long-distance targets and large fields of view, emits electromagnetic waves at a certain frequency, reflects the electromagnetic waves after encountering an object, and detects the reflected electromagnetic waves at a gap between the radar and the non-emitted electromagnetic waves, so that positioning and tracking are realized. With the development of various image sensors and components, compared with a radar system, the image-based target detection tracking system has relatively low cost and wider application range.
Image-based moving object detection tracking and imaging are two mutually independent parts, namely, the moving object needs to be positioned from continuous clear images to detect tracking, and the accuracy depends on the quality of the captured images. In this case, the factors affecting the image quality are mainly motion blur caused by a fast moving object, and in order to cope with the motion blur, an imaging system and an image processing related algorithm with high time resolution are required, firstly, for a high-speed camera with high time resolution, it is expensive, and in a very short time, the data throughput is usually large, a large data storage capacity and a wide data transmission bandwidth are required, and secondly, the advanced image processing and analysis algorithm also occupies a large amount of calculation resources. In some existing technologies, high refresh rate and efficient reconstruction algorithms based on image moving object detection and tracking are discussed, if the moving speed of the object is too high, these methods are not applicable, and some technical methods need a high-precision tracking and aiming system for acquiring the moving information through the evolution process of the moving object, propose that the object tracking is not needed by clear images, and are obtained by the interrelation between rough images of the object at different positions, and imaging is also carried out substantially, so that the consumption of time and space costs still exists in the imaging process.
Therefore, the image-based method has the advantages that huge time and space expenses are brought to the acquisition of a large amount of image data for target detection and tracking, the resource waste is caused, the redundancy of data acquisition is one of main reasons for real-time long-time target detection and tracking, and in practical application, the complex image processing analysis algorithm and the limited data processing capacity bring challenges to real-time target detection and tracking.
Disclosure of Invention
The technical purpose is that: aiming at the technical problems, the invention provides a non-imaging real-time detection tracking method and a non-imaging real-time detection tracking system for a fast moving object, which do not need to acquire a target scene image, acquire key information of a target space position by using a spatial light modulator and a single pixel detector, have ultralow sampling rate and higher calculation efficiency, and can realize real-time detection and tracking of a fast target.
The technical scheme is as follows: in order to achieve the technical purpose, the invention adopts the following technical scheme:
a non-imaging real-time detection tracking method for a fast moving object comprises the following steps:
s1, constructing a Hadamard matrix M according to the scene size, and remolding each row or each column of the Hadamard matrix M to obtain a modulation base sequence H;
s2, optimizing the modulation base sequence H, wherein the optimized modulation base sequence H comprises a modulation base for realizing the high concentrated image energy of the preferential projection energy;
s3, decomposing the optimized modulation base sequence H according to rows and columns in the X-axis direction and the Y-axis direction, and reconstructing each decomposed row and each decomposed column to obtain a sub-modulation base sequence SubPattern;
s4, before the fast moving object enters the scene, projecting the sub-modulation base sequence SubPattern into the scene, reconstructing to obtain a projection curve without a moving object in the X-axis or Y-axis direction by the projection value of the sub-modulation base and the measured value acquired by the single pixel detector, taking the projection curve as a priori projection curve, and calculating the first-order gradient of the priori projection curve;
s5, starting from the fast moving object entering the scene, projecting the sub-modulation base sequence SubPattern into the scene, and reconstructing to obtain a projection curve with a moving object in the X-axis or Y-axis direction through the projection value of the sub-modulation base and the measured value acquired by the single-pixel detector;
s6, calculating a first-order gradient of a projection curve of the moving object;
s7, calculating to obtain gradient differences in the X-axis and Y-axis directions according to the first-order gradient of the prior projection curve and the first-order gradient of the projection curve with the moving object;
and S8, extracting the first two maximum values of the gradient difference in the X-axis direction and the Y-axis direction to obtain the position information of the fast moving object in the X-axis direction and the Y-axis direction, and further determining the position of the fast moving object in the scene.
Preferably, in the step S1, the scene size is S 1 ×S 1 The length and width of the constructed Hadamard matrix M are S 1 2 ×S 1 2 S is present in the desired modulation base sequence H 1 2 A plurality of modulation bases, each modulation base having a size S 1 ×S 1 。
Preferably, in the step S3, the sub-modulation base sequence SubPattern is obtained by decomposing the rows and columns of the modulation base sequence H, and the formula (1) is as follows:
SubPattern[i](1:S 1 ,:)＝H[j](u,:) (1)
wherein SubPattern [ i ]]Represents the ith sub-modulation base, H [ j ] in the sub-modulation base sequence SubPattern]Represents the j-th modulation base in the modulation base sequence H, and formula (1) represents the 1 st line to S of the i-th sub-modulation base in the sub-modulation base sequence SubPattern 1 Lines, each line being equal to the j-th modulation base H [ j ] in the modulation base sequence H]Is the u th line of (2).
Preferably, in the step S4:
reconstructing a projection curve F of the scene without moving targets according to the formula (2) o (i)，
F o (i)＝IP o (P,I)*C o (i),i＝1,2,…,S 1 (2)
Calculating a motionless object scene projection curve F using the previous element minus the next element in the vector according to equation (3) o (i) First order gradient F' o (i)，
Wherein, IP o (P, I) represents the measured value acquired by the single-pixel detector in the scene without moving targets, C o (i) Representing the projection of the sub-modulation base in the X-axis direction or the Y-axis direction in the scene without moving targets.
Preferably, in the step S5, a projection curve F of the moving object scene is reconstructed according to the formula (4) M (i)：
F M (i)＝IP M (P,I)*C M (i),i＝1,2,…,S 1 (4)
In the step S6, a first-order gradient F 'of the projection curve of the scene with the moving object is calculated by subtracting the latter element from the former element in the vector according to the formula (5)' M (i)：
Wherein, IP M (P, I) represents the measured value acquired by the single-pixel detector in the scene with the moving object, C M (i) The projection of the sub-modulation base in the X-axis direction or the Y-axis direction in the scene with the moving object is shown.
Preferably, in the step S7, a difference operation is performed between the first-order gradient of the prior projection curve and the first-order gradient of the scene projection curve after the moving object enters according to the formula (6):
wherein Diff represents the differential operation,representing a step difference result.
Preferably, in the step S8, positional information of the fast moving object, i.e., the moving object, in the X-axis and Y-axis directions is determined according to the formula (7) and the formula (8):
wherein, (v) x1 ,v x2 ) Boundary values representing the X-axis direction of the object in the scene, (v) y1 ,v y2 ) Representing the boundary value of the object in the Y-axis direction in the scene, slt represents the maximum two values in the selection vector,representing the result comprising a step difference on the X-axis, -/->The representation includes a one-step differential result on the Y-axis.
A fast moving object non-imaging real-time detection tracking system comprising: the system comprises a light source, a spatial light modulator, a single-pixel detector, a data acquisition unit and a computer, wherein the computer is provided with a memory and a processor, the memory stores a computer program, the computer program is configured to realize the steps of the method when being called by the processor, the spatial light modulator is used for receiving a sub-modulation base designed by the computer, receiving light rays irradiated by the light source and projecting the light rays into a scene according to an image of the sub-modulation base, a scene panel in the scene and a light intensity signal response value on a target object entering a scene panel area are measured through the single-pixel detector, and data measured by the single-pixel detector are acquired through the data acquisition unit and sent to the computer.
The beneficial effects are that: due to the adoption of the technical scheme, the invention has the following beneficial effects:
the invention discloses an image-free real-time moving target detection method and system based on projection curve gradient difference, which realize real-time detection and tracking of a fast moving target in a continuous scene with a complex background.
Drawings
FIG. 1 is a flow chart of a fast moving object non-imaging real-time detection tracking method according to the present invention;
FIG. 2 is a schematic diagram of a Hadamard matrix;
FIG. 3 is a modulation based image;
FIG. 4 is a reconstructed sub-modulation basis image;
FIG. 5 is a schematic view of a projected curve of a non-moving object scene in the X-axis direction and the Y-axis direction;
FIG. 6 is a schematic diagram of a first order gradient of the non-moving object scene projection curve of FIG. 5;
FIG. 7 is a schematic view of projection curves of a moving object scene in the X-axis direction and the Y-axis direction;
FIG. 8 is a schematic diagram of a first order gradient of the moving object scene projection curve of FIG. 7;
FIG. 9 is a one-step difference effect plot of a non-moving object scene versus a moving object scene projection curve;
fig. 10 is a schematic diagram of optical system imaging.
Detailed Description
Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
The invention discloses a projection curve gradient difference-based non-imaging real-time detection and tracking scheme for a moving target, which is used for realizing real-time detection and tracking of a fast moving target in a continuous scene with a complex background. The method utilizes a spatial light modulator and a single-pixel detector to acquire key information of a target spatial position under the condition that the target scene image is not required to be acquired. In addition, the invention adopts an optimized Hadamard (Hadamard) modulation base sequence, uses a small amount of modulation base Pattern to modulate an object, carries out signal measurement, realizes ultralow sampling rate, then utilizes a single-pixel detector to collect light intensity signals, converts a calculation object from a two-dimensional image to a one-dimensional space, reduces the calculation amount, improves the real-time performance of detection and tracking, directly adopts a single-pixel measurement value to reconstruct the position information of the target object under the condition of no scene image, and realizes real-time detection and tracking of a fast moving target.
Example 1
As shown in fig. 1, the present embodiment discloses a fast moving object non-imaging real-time detection and tracking method based on a Single-Pixel Detector (Single-Pixel Detector), which includes the following steps:
step 1: s according to the required scene size 1 ×S 1 Constructing a Hadamard (Hadamard) matrix with a length and a width of S 1 2 ×S 1 2 The method comprises the steps of carrying out a first treatment on the surface of the Remolding each row or each column of the Hadamard matrix to obtain a Hadamard modulation base sequence H, wherein S is arranged in the sequence 1 2 A plurality of modulation bases, each modulation base having a size S 1 ×S 1 。
Step 2: the modulation base sequence H is optimized, the purpose of optimization is to realize the preferential projection of the modulation base (namely, a small number of coefficients can acquire larger energy) with high concentrated image energy, and the rest of the modulation base sequence H is ignored, so that the information acquisition efficiency is improved; the projection order can be optimized by modulating the area size of the active area (white area) of the base, the larger the active area is, the more the modulation base can concentrate the energy of the image. Each modulation base corresponds to one coefficient, and the efficient acquisition of information can be realized by adopting a small number of modulation bases, namely, the fact that a small number of coefficients have larger assignment capability is reflected.
Step 3: in the X-axis direction and the Y-axis direction, the Hadamard modulation bases are decomposed in rows and columns, and each row and each column of the decomposed Hadamard modulation bases are reconstructed to obtain a sub-modulation base sequence SubPattern, subPattern to be projected to a scene containing a fast moving object.
Step 4: in order to reduce the amount of computation, the projection curve is reconstructed for a non-moving object scene before a fast moving object enters using the principle that the light intensity measurement is mathematically equivalent to modulating the inner product between the base image and the object, and its first order gradient is calculated as a reference term.
Step 5: starting from the entry of a fast moving object into the scene, projection curves in the X-axis direction and in the Y-axis direction are reconstructed using the principle that the light intensity measurement is mathematically equivalent to the inner product between the modulated base image and the object.
Step 6: a first order gradient of a projection curve of the scene (into which the moving object has entered) is calculated.
Step 7: a gradient difference operation is performed on the first order gradient between the scene and the previous projection curve (reference term).
Step 8: and extracting the first two maximum values of the gradient difference result to obtain the position information of the object in the X-axis or Y-axis direction in the scene.
Example 1:
step 1: s according to the required scene size 1 ×S 1 Constructing a Hadamard (Hadamard) matrix M with a length and a width of S 1 2 ×S 1 2 The method comprises the steps of carrying out a first treatment on the surface of the Remolding each row or each column of the Hadamard matrix to obtain a Hadamard modulation base sequence H, wherein S is arranged in the sequence 1 2 A plurality of modulation bases, each modulation base having a size S 1 ×S 1 ；
S1.1, in the process of algorithm verification, targets in a scene with the size of 128 multiplied by 128 are detected and tracked, so that the length of each row or each column of the constructed Hadamard matrix is 16384, and the size of the Hadamard matrix is 16384 multiplied by 16384, which is shown in the attached figure 2.
S1.2, remolding each row or each column of the Hadamard matrix to obtain a required modulation base sequence H;
s1.3, in the verification process of the invention, detecting and tracking targets in a scene with the size of 128×128, taking each row in a Hadamard matrix for remodeling, namely, vector remodeling with the size of 1×16384, obtaining a modulation base image with the size of 128×128, wherein 16384 rows are included in the obtained modulation base sequence, and the modulation base image is shown in figure 3.
Step 2: the modulation base sequence is optimized by using the existing Hadamard (Hadamard) modulation base sequence optimization method, so that the modulation base which can concentrate the image energy highly is projected preferentially (namely, a small number of coefficients have larger assignment), the rest of the modulation base is ignored, and the information acquisition efficiency is improved;
step 3: decomposing the Hadamard modulation bases in the X-axis direction and the Y-axis direction according to rows and columns respectively, and reconstructing each decomposed row and each decomposed column to obtain a sub-modulation base sequence SubPattern, subPattern to be projected to a scene containing a fast moving object;
s3.1, obtaining a sub-modulation base sequence by decomposing rows and columns of the modulation base, wherein the formula is as follows:
SubPattern[i](1:128,:)＝H[j](u,:)
wherein SubPattern [ i ] represents the ith sub-modulation base in the sequence, H [ j ] represents the jth modulation base in the H sequence, and the formula represents the ith sub-modulation base in the (1 st) row to the (128 th) row, each row being equal to H [ j ], see FIG. 4.
Step 4: in order to reduce the calculation amount, the principle that the light intensity measurement is mathematically equivalent to the inner product between the modulation base image and the object is utilized, a projection curve (prior projection curve) is reconstructed for a scene without a moving object before the fast moving object enters, and the first-order gradient of the projection curve is used as a reference item;
s4.1, an inner product formula between the sub-modulation base image and the object is as follows:
wherein P represents a sub-modulation base, I represents a target scene, and the coordinate system is a Cartesian coordinate system; p (x, y) represents a sub-modulation base image, the upper left corner of the image is taken as an origin (0, 0) of coordinates, and (x, y) represents the coordinates of a certain pixel point in the image; the target scene I (x, y) is the same. The intensity of illumination obtained at the projection of the sub-modulation base onto the target scene is measured by a single-pixel detector, the measured value being approximately equal to IP.
S4.2, reconstructing a projection curve of the non-moving object scene in the X-axis direction and the Y-axis direction, wherein the specific formula principle is as follows:
PC x with PC (personal computer) y The vector is a 1×128 vector, which represents the projection of the sub-modulation base in the X-axis direction and the Y-axis direction, and is obtained by further decomposing and reconstructing P (X, Y) in the X-axis direction and the Y-axis direction on the basis of P (X, Y), which is shown in fig. 4. The overall representation of C (i) may be used as follows:
C(i)＝PC(i),i＝1,2,…,128.
reconstructing a projection curve of the scene by the projection value of the sub-modulation base and the measured value acquired by the single-pixel detector, wherein the projection curve comprises the following formula:
F(i)＝IP(P,I)*C(i),i＝1,2,…,128.
f represents the reconstructed projection curve, which is a 1×128 vector, see fig. 5. The method adopts the optimized two-dimensional matrix to ensure that the modulation bases capable of acquiring more information are projected preferentially after projection, thereby reducing the use of the number of the modulation bases and improving the efficiency.
S4.3, subtracting the latter element from the former element in the vector to calculate a first-order gradient of the projection curve F, wherein the formula is as follows:
the invention uses F' o (i) A first order gradient representing the a priori projection curve is used as a subsequent reference term, as follows:
the first-order gradient chart is shown in fig. 6. In the method, the light intensity measurement is directly obtained through the single-pixel detector in actual operation, and inner product calculation is not needed. In the simulation process, the inner product operation of the object image and the modulation base image is used to approximate the direct measurement of light intensity.
Step 5: reconstructing projection curves in the X-axis direction and the Y-axis direction from the entry of the fast moving object into the scene using a principle that the light intensity measurement is mathematically equivalent to the inner product between the modulated base image and the object;
s5.1, after a moving object enters a scene, reconstructing a projection curve of the scene, wherein the principle is consistent with that described in S4.1 and S4.2, and the effect diagram is shown in the attached figure 7;
step 6: calculating a first-order gradient of a projection curve of a scene (a moving object enters the scene);
s6.1, the invention uses F' M (i) A first order gradient representing a projection curve of a moving object after entering a scene is as follows:
the first-order gradient chart is shown in fig. 8.
Step 7: performing gradient difference operation on a first-order gradient between a scene and a priori projection curve (reference item); according to the method, the first-order gradient is calculated, the edges of the moving object and the scene can be extracted, and then the boundary coordinate information of the object, namely the position information of the object in the scene, is obtained.
S7.1, performing differential operation on the first-order gradient of the prior projection curve and the first-order gradient of the scene projection curve after the moving object enters, so as to obtain the position information of the object, wherein the differential operation formula is as follows:
Diff[m(i),n(i)]＝abs[m(i),n(i)],i＝1,2,…,128.
thus, a step difference operation is as follows:
wherein Diff represents the differential operation,representing a step difference result.
The calculation mode of the position information in the X-axis direction and the Y-axis direction is identical, so that the principle derivation is only performed once here.
Step 8: extracting the first two maximum values of the gradient difference result to obtain the position information of the object in the X-axis or Y-axis direction in the scene;
S8.1、is a vector, extracts maximum two values, and can obtain corresponding position information in X-axis or Y-axis direction, using +.>The representation contains position information on the X-axis, < >>The representation includes position information on the Y-axis, and there are:
wherein (v) x1 ,v x2 ) Boundary values representing the X-axis direction of the object in the scene, (v) y1 ,v y2 ) Boundary values representing the Y-axis direction of the object in the scene, slt represents the selection vectorThe graph effect is shown in figure 9.
From the above, the method of the present invention realizes real-time detection and tracking of a fast moving object in a continuous scene with a complex background, and the method realizes acquisition of key information of a target spatial position by using a spatial light modulator and a single-pixel detector without acquisition of an image of the target scene. The invention can realize the detection and tracking of the fast moving object in the continuous scene, essentially by acquiring the two-dimensional or three-dimensional space coordinate information (Cartesian coordinate system) of the object in the scene, the key space information of the object can be determined by only two or three scalar quantities irrespective of the attribute of the object, and compared with the imaging of thousands of pixel information (each pixel has 2 scalar quantities for representation), the calculated quantity is reduced by 3 to 4 orders of magnitude, thus solving the problems of huge time and space expenditure existing in the image-based method.
Example two
As shown in fig. 10, the present embodiment provides a fast moving object non-imaging real-time detection system based on a single pixel detector, including: an LED light source, a Collecting lens (DMD), a Digital Micromirror (DMD), a projection lens (projection lens), a Scene panel (Scene plane), a moving object/fast moving object (Target object), a PAD provided with a single pixel detector, a data collector DAQ, and a computer PC.
Wherein the computer PC is provided with a memory and a processor, the memory storing a computer program configured to, when called by said processor, carry out the steps of the method described in embodiment one.
The digital micro-mirror DMD is used as a spatial light modulator, receives a sub-modulation base designed by the computer PC, modulates light rays passing through the collecting lens according to an image of the sub-modulation base, irradiates a scene panel through the projection lens, and detects light intensity signal response values on the scene panel and a target object entering the scene panel area by a single-pixel detector on the PAD, and data detected by the single-pixel detector are collected by the DAQ and sent to the computer PC.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be appreciated by persons skilled in the art that the above embodiments are not intended to limit the invention in any way, and that all technical solutions obtained by means of equivalent substitutions or equivalent transformations fall within the scope of the invention.
Claims (8)
1. The non-imaging real-time detection tracking method for the fast moving object is characterized by comprising the following steps:
s1, constructing a Hadamard matrix M according to the scene size, and remolding each row or each column of the Hadamard matrix M to obtain a modulation base sequence H;
s2, optimizing the modulation base sequence H, wherein the optimized modulation base sequence H comprises a modulation base for realizing the high concentrated image energy of the preferential projection energy;
s3, decomposing the optimized modulation base sequence H according to rows and columns in the X-axis direction and the Y-axis direction, and reconstructing each decomposed row and each decomposed column to obtain a sub-modulation base sequence SubPattern;
s4, before the fast moving object enters the scene, projecting the sub-modulation base sequence SubPattern into the scene, reconstructing to obtain a projection curve without a moving object in the X-axis or Y-axis direction by the projection value of the sub-modulation base and the measured value acquired by the single pixel detector, taking the projection curve as a priori projection curve, and calculating the first-order gradient of the priori projection curve;
s5, starting from the fast moving object entering the scene, projecting the sub-modulation base sequence SubPattern into the scene, and reconstructing to obtain a projection curve with a moving object in the X-axis or Y-axis direction through the projection value of the sub-modulation base and the measured value acquired by the single-pixel detector;
s6, calculating a first-order gradient of a projection curve of the moving object;
s7, calculating to obtain gradient differences in the X-axis and Y-axis directions according to the first-order gradient of the prior projection curve and the first-order gradient of the projection curve with the moving object;
and S8, extracting the first two maximum values of the gradient difference in the X-axis direction and the Y-axis direction to obtain the position information of the fast moving object in the X-axis direction and the Y-axis direction, and further determining the position of the fast moving object in the scene.
2. The method for non-imaging real-time detection and tracking of a fast moving object according to claim 1, wherein the method comprises the steps of: in the step S1, the scene size is S 1 ×S 1 The length and width of the constructed Hadamard matrix M are S 1 2 ×S 1 2 S is present in the desired modulation base sequence H 1 2 A plurality of modulation bases, each modulation base having a size S 1 ×S 1 。
3. The method for non-imaging real-time detection and tracking of a fast moving object according to claim 2, wherein: in the step S3, the sub-modulation base sequence SubPattern is obtained by decomposing the rows and columns of the modulation base sequence H, and the formula (1) is as follows:
SubPattern[i](1：S 1 ，：)＝H[j](u，：) (1)
wherein SubPattern [ i ]]Represents the ith sub-modulation base, H [ j ] in the sub-modulation base sequence SubPattern]Represents the j-th modulation base in the modulation base sequence H, and formula (1) represents the 1 st line to S of the i-th sub-modulation base in the sub-modulation base sequence SubPattern 1 Lines, each line being equal to the j-th modulation base H [ j ] in the modulation base sequence H]Is the u th line of (2).
4. The fast moving object non-imaging real-time detection tracker according to claim 2, wherein in step S4:
reconstructing a projection curve F of the scene without moving targets according to the formula (2) o (i)，
F o (i)＝IP o (P，I)*C o (i)，i＝1，2，…，S 1 (2)
Calculating a motionless object scene projection curve F using the previous element minus the next element in the vector according to equation (3) o (i) First order gradient F' o (i)，
Wherein, IP o (P, I) represents the measured value acquired by the single-pixel detector in the scene without moving targets, C o (i) Representing the projection of the sub-modulation base in the x-axis direction or the Y-axis direction in the scene without moving targets.
5. The method according to claim 4, wherein in step S5, the projection curve F of the moving object scene is reconstructed according to the formula (4) M (i)：
F M (i)＝IP M (P，I)*C M (i)，i＝1，2，…，S 1 (4)
In the step S6, a first-order gradient F 'of the projection curve of the scene with the moving object is calculated by subtracting the latter element from the former element in the vector according to the formula (5)' M (i)：
Wherein, IP M (P, I) represents the measured value acquired by the single-pixel detector in the scene with the moving object, C M (i) The projection of the sub-modulation base in the X-axis direction or the Y-axis direction in the scene with the moving object is shown.
6. The method according to claim 5, wherein in step S7, a difference operation is performed between a first-order gradient of the prior projection curve and a first-order gradient of the scene projection curve after the moving object enters according to equation (6):
wherein Diff represents the differential operation,representing a step difference result.
7. The method according to claim 6, wherein in the step S8, the positional information of the fast moving object, i.e., the moving object, in the X-axis and Y-axis directions is determined according to the formulas (7) and (8):
wherein, (v) x1 ，v x2 ) Boundary value (v) representing X-axis direction of moving object in scene y1 ，v y2 ) Representing the boundary value of the object in the Y-axis direction in the scene, slt represents the maximum two values in the selection vector,representing the result comprising a step difference on the X-axis, -/->The representation includes a one-step differential result on the Y-axis.
8. A fast moving object non-imaging real-time detection tracking system, comprising: the system comprises a light source, a spatial light modulator, a single-pixel detector, a data acquisition unit and a computer, wherein the computer is provided with a memory and a processor, the memory stores a computer program, the computer program is configured to realize the steps of the method of any one of claims 1 to 7 when the computer program is called by the processor, the spatial light modulator is used for receiving a sub-modulation base designed by the computer, receiving light rays irradiated by the light source and projecting the light rays into a scene according to an image of the sub-modulation base, a scene panel in the scene and a light intensity signal response value on a target object entering a scene panel area are measured by the single-pixel detector, and data measured by the single-pixel detector are acquired by the data acquisition unit and sent to the computer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311164191.0A CN117132622A (en) | 2023-09-08 | 2023-09-08 | Non-imaging real-time detection tracking method and system for fast moving object |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311164191.0A CN117132622A (en) | 2023-09-08 | 2023-09-08 | Non-imaging real-time detection tracking method and system for fast moving object |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117132622A true CN117132622A (en) | 2023-11-28 |
Family
ID=88856348
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311164191.0A Pending CN117132622A (en) | 2023-09-08 | 2023-09-08 | Non-imaging real-time detection tracking method and system for fast moving object |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117132622A (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108895985A (en) * | 2018-06-19 | 2018-11-27 | 中国科学院合肥物质科学研究院 | A kind of object positioning method based on single pixel detector |
| CN116125492A (en) * | 2022-12-15 | 2023-05-16 | 中国科学技术大学 | Ultra-high-speed single-pixel imaging device and method |
-
2023
- 2023-09-08 CN CN202311164191.0A patent/CN117132622A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108895985A (en) * | 2018-06-19 | 2018-11-27 | 中国科学院合肥物质科学研究院 | A kind of object positioning method based on single pixel detector |
| CN116125492A (en) * | 2022-12-15 | 2023-05-16 | 中国科学技术大学 | Ultra-high-speed single-pixel imaging device and method |
Non-Patent Citations (1)
| Title |
|---|
| FENGMING ZHOU: "Non-imaging real-time detection and tracking of fast-moving objects using a single-pixel detector", 《HTTPS://ARXIV.ORG/ABS/2108.06009》, pages 1 - 16 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11151739B2 (en) | Simultaneous localization and mapping with an event camera | |
| Rebecq et al. | EMVS: Event-based multi-view stereo—3D reconstruction with an event camera in real-time | |
| US8848035B2 (en) | Device for generating three dimensional surface models of moving objects | |
| US20200057831A1 (en) | Real-time generation of synthetic data from multi-shot structured light sensors for three-dimensional object pose estimation | |
| JP2012517651A (en) | Registration of 3D point cloud data for 2D electro-optic image data | |
| CN110349186B (en) | Large-displacement motion optical flow calculation method based on depth matching | |
| CN107203743B (en) | Face depth tracking device and implementation method | |
| Zheng et al. | Deep learning for event-based vision: A comprehensive survey and benchmarks | |
| EP2476999B1 (en) | Method for measuring displacement, device for measuring displacement, and program for measuring displacement | |
| CN110097634B (en) | Self-adaptive multi-scale three-dimensional ghost imaging method | |
| CN113610889A (en) | Human body three-dimensional model obtaining method and device, intelligent terminal and storage medium | |
| US20230394833A1 (en) | Method, system and computer readable media for object detection coverage estimation | |
| Khan et al. | High-density single shot 3D sensing using adaptable speckle projection system with varying preprocessing | |
| CN111596310B (en) | Moving target ghost imaging system and method based on point detection | |
| Ouyang et al. | Dynamic depth fusion and transformation for monocular 3d object detection | |
| US8818124B1 (en) | Methods, apparatus, and systems for super resolution of LIDAR data sets | |
| JPH04130587A (en) | Three-dimensional picture evaluation device | |
| CN117132622A (en) | Non-imaging real-time detection tracking method and system for fast moving object | |
| Loktev et al. | Image Blur Simulation for the Estimation of the Behavior of Real Objects by Monitoring Systems. | |
| JPH11248431A (en) | Three-dimensional model forming apparatus and computer readable medium recorded with three-dimensional model generating program | |
| KR20120056668A (en) | Apparatus and method for recovering 3 dimensional information | |
| Subpa-Asa et al. | Separating the direct and global components of a single image | |
| CN117079117B (en) | Underwater image processing and target identification method and device, storage medium and electronic equipment | |
| Zhou | Non-imaging real-time detection and tracking of fast-moving objects using a single-pixel detector | |
| CN113192154B (en) | Underwater ghost imaging system based on edge calculation and deep learning image reconstruction method |
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 | ||
| CB02 | Change of applicant information |
Address after: 210000 Room 301, floor 3, building 75, zone B, entrepreneurship and innovation city, No. 15, Fengji Avenue, Yuhuatai District, Nanjing, Jiangsu Province Applicant after: Nanjing Thunderbolt Information Technology Co.,Ltd. Address before: 210000 Room 301, floor 3, building 75, zone B, entrepreneurship and innovation city, No. 15, Fengji Avenue, Yuhuatai District, Nanjing, Jiangsu Province Applicant before: NANJING LEADING INFORMATION TECHNOLOGY Co.,Ltd. |
|
| CB02 | Change of applicant information |