CN-121980327-A - Detection and identification method based on photo-magnetic fusion

CN121980327ACN 121980327 ACN121980327 ACN 121980327ACN-121980327-A

Abstract

The invention designs a detection and identification method based on photo-magnetic fusion, and belongs to the technical field of target detection and identification. And (3) carrying optical imaging equipment and magnetic measurement load on the motion platform, collecting optical image data and magnetic measurement data in a task area, performing enhancement processing and target detection on the optical image data, and extracting surface target information. And performing background magnetic field elimination and noise suppression on the magnetic measurement data, extracting magnetic anomaly data and generating a magnetic anomaly distribution map. And superposing the magnetic anomaly distribution map and the optical image, performing space matching and object classification, performing magnetic field modeling on the visible ferromagnetic object, and stripping the contribution from the total magnetic anomaly to enhance the signal of the hidden object. And carrying out target identification through the classification model, and outputting the target type, the position profile and the magnetic characteristic parameters. The invention combines optical and magnetic anomaly detection technologies, improves the sensitivity and resolution of target detection in complex environments, and is suitable for the fields of military reconnaissance, geological exploration, security monitoring and the like.

Inventors

SHI YUFENG
XIAO TING
SONG XIAOLIN
WANG WENCONG
Ren wenguan
Qu yanan
SONG XIAOLIN

Assignees

山东航天电子技术研究所

Dates

Publication Date: 20260505
Application Date: 20251210

Claims (10)

1. The detection and identification method based on photo-magnetic fusion is characterized by comprising the following steps of: S1, acquiring data, namely carrying optical imaging equipment and magnetic measurement load by using a motion platform, acquiring optical image data and magnetic measurement data according to a planning and measuring line in a task area, wherein the magnetic measurement data comprises measurement position data and measurement height data; S2, optical data processing, namely performing image enhancement processing on the optical image data to generate an enhanced optical image, extracting surface targets in the enhanced optical image by using a target detection algorithm, and marking the type and position information of each surface target to generate an optical target detection result; S3, processing magnetic anomaly data, namely performing background magnetic field elimination and noise suppression processing on the magnetic measurement data, extracting the magnetic anomaly data, then acquiring position and contour information of a magnetic anomaly target by utilizing a target positioning algorithm, and then converting the magnetic anomaly data into regular grid data through grid interpolation processing to generate a magnetic anomaly distribution map; S4, photo-magnetic data fusion processing, namely superposing the magnetic anomaly distribution map on the enhanced optical image, performing space matching on the optical target detection result and the position of the magnetic anomaly target, performing preliminary classification on the target according to the matching result, performing magnetic field modeling on the visible ferromagnetic target with larger magnetic anomaly intensity, calculating magnetic anomaly distribution generated in a task area, subtracting the magnetic anomaly distribution generated by the visible ferromagnetic target from total magnetic anomaly data of the task area to obtain residual magnetic anomaly data, and enhancing magnetic anomaly signals of a hidden target; S5, target identification and output, namely inputting the data subjected to the photomagnetic fusion into a classification model for target identification, and outputting the target type, the target position profile and the magnetic characteristic parameters.
2. The method of claim 1, wherein the step S1 comprises the sub-steps of: S11, line planning, namely setting line height and line interval parameters according to a task area range, flight safety requirements and resolution requirements, and planning a flight line of the motion platform; S12, data acquisition, namely controlling the motion platform to fly according to a planned flight survey line, and synchronously or time-sharing data acquisition is carried out on the optical imaging equipment and the magnetic survey load to acquire optical image data and magnetic survey data containing a time stamp; S13, spatial registration, namely performing spatial alignment on the optical image data and the magnetic measurement data based on a registration method of characteristic point matching, wherein the spatial registration comprises the steps of extracting characteristic points from the optical image, establishing a corresponding relation between magnetic measurement points and optical image pixels according to position data in the magnetic measurement data, realizing spatial registration of the two types of data through affine transformation or perspective transformation, and generating a multi-mode dataset after spatial alignment.
3. The method of claim 1, wherein the step S2 comprises the sub-steps of: s21, image preprocessing, namely denoising the optical image data to eliminate random noise introduced in the image acquisition process; S22, performing contrast enhancement on the denoised optical image by adopting a self-adaptive histogram equalization method, dividing the image into a plurality of subareas, and respectively performing histogram equalization processing on each subarea to generate an enhanced optical image; s23, target detection, namely performing target detection on the enhanced optical image by using a deep learning target detection model or an edge detection algorithm, and identifying surface targets such as surface buildings, vehicles, bridges, roads and the like; And S24, numbering each detected surface target, and recording the type name, the central position coordinate and the outline boundary box of each detected surface target to generate an optical target detection result.
4. The method of claim 1, wherein the step S3 comprises the sub-steps of: S31, eliminating a background magnetic field, namely eliminating the influence of the earth background magnetic field by adopting a trending method according to the sampling rate of the magnetic measurement data and the flying speed of the motion platform, and extracting a local magnetic anomaly signal; s32, noise suppression, namely filtering the local magnetic anomaly signal by adopting wavelet transformation, kalman filtering or self-adaptive filtering methods, eliminating environmental magnetic noise and acquiring magnetic anomaly data; s33, target positioning, namely processing the magnetic anomaly data by adopting an analytic signal method, an Euler inversion method or a matched filtering method to acquire horizontal position coordinates and depth information of the magnetic anomaly target; S34, gridding interpolation, namely selecting a Kriging interpolation method, an inverse distance weighted interpolation method or a spline interpolation method according to the density and the survey line type of the magnetic survey data, converting irregularly-distributed magnetic anomaly data into regular grid data, and generating a magnetic anomaly distribution map.
5. The method for detecting and identifying based on photo-magnetic fusion according to claim 1, wherein the step S4 of primarily classifying the targets according to the matching result comprises the following sub-steps: S41, performing image superposition, namely performing superposition display on the magnetic anomaly distribution map and the enhanced optical image according to the same space coordinate system to generate a magneto-optical superposition image; S42, space matching, namely traversing each earth surface target in the optical target detection result, detecting whether a magnetic abnormal signal exists at the position of each earth surface target, traversing each magnetic abnormal region in the magnetic abnormal distribution diagram, and detecting whether an optical visible target exists at the position of each earth surface target; s43, primarily classifying the targets according to the space matching result, wherein the targets are classified into three types, namely camouflage targets, namely targets visible in the optical image and having no magnetic abnormal response, visible ferromagnetic targets, namely targets visible in the optical image and the magnetic abnormal distribution map, and camouflage targets, namely targets invisible in the optical image and having magnetic abnormal response.
6. The detection and recognition method based on photo-magnetic fusion according to claim 5, wherein in the step S4, the visible ferromagnetic target with larger magnetic anomaly strength is subjected to magnetic field modeling, and the magnetic anomaly distribution generated in the task area is calculated, and the method comprises the following substeps: S44, screening interference targets, namely screening targets with abnormal magnetic strength larger than a preset threshold value from the visible ferromagnetic targets as magnetic interference targets; s45, extracting magnetic anomaly data in each magnetic interference target contour area; S46, magnetic field modeling, namely establishing a magnetic field model for each magnetic interference target by adopting a multi-magnetic dipole equivalent method, and determining the position, the magnetic moment and the direction parameters of each magnetic dipole by least square fitting; S47, calculating magnetic anomaly values generated by each magnetic interference target at grid points in the whole task area according to the established magnetic field model, and accumulating the magnetic anomaly values generated by all the magnetic interference targets to obtain total magnetic anomaly distribution of the magnetic interference targets; S48, stripping the magnetic anomaly, namely subtracting the total magnetic anomaly distribution of the magnetic interference targets from the total magnetic anomaly data of the task area to obtain residual magnetic anomaly data, and enhancing the magnetic anomaly signal visualization effect of the weak magnetic targets and the hidden targets.
7. The detection and recognition method based on photo-magnetic fusion according to claim 6, wherein in the step S46, a magnetic field model is built for a magnetic interference target by adopting a multi-magnetic dipole equivalent method, and the specific implementation process is as follows: S461, initializing magnetic dipole parameters, namely determining the quantity of equivalent magnetic dipoles according to the outline size of the magnetic interference target Uniformly arranged within the target contour region Initial positions of the magnetic dipoles; s462, constructing a magnetic field forward model, namely calculating a magnetic anomaly theoretical value generated by each magnetic dipole at a measuring point according to a magnetic field formula of the magnetic dipoles, wherein the total field intensity of the magnetic anomaly generated by each magnetic dipole at any point in space is calculated by the following formula: ; Wherein, the A magnetic anomaly (nT) generated at the observation point for the magnetic dipole; Is vacuum magnetic permeability, takes the value of (H/m); A magnetic moment (A.m2) which is a magnetic dipole; the distance (m) from the observation point to the magnetic dipole; An included angle (rad) between the direction of the observation point and the direction of the magnetic moment; Is the circumference ratio; And 463, parameter inversion, namely using the minimum sum of squares of residual errors between actually measured magnetic anomaly data and theoretical calculated values as an objective function, and adopting a least square method to iteratively optimize the position coordinates and magnetic moment parameters of each magnetic dipole until the residual errors are converged to a preset precision, so as to complete the establishment of a magnetic field model.
8. The method for detecting and identifying based on photo-magnetic fusion according to claim 1, wherein the step S5 comprises the following sub-steps: s51, extracting characteristics, namely extracting target characteristics from the data subjected to photomagnetic fusion, wherein the target characteristics comprise optical characteristics and magnetic characteristics, the optical characteristics comprise target shape, texture and color information, and the magnetic characteristics comprise magnetic anomaly amplitude, magnetic anomaly range and magnetic anomaly gradient information; S52, classifying and identifying, namely inputting the extracted target features into a pre-trained classification model, wherein the classification model is a convolutional neural network or a support vector machine, and outputting a type label and a confidence score of the target; and S53, outputting a result, namely outputting a target with the opposite confidence score higher than a preset threshold value, and outputting a target type, position profile information and magnetic characteristic parameters of the target, wherein the target type comprises underground work, underground pipelines, camouflage vehicles and surface buildings, and the magnetic characteristic parameters comprise equivalent magnetic moment and burial depth information.
9. The detection and recognition method based on photo-magnetic fusion according to claim 3, wherein the step S22 is implemented by performing contrast enhancement on the image by using an adaptive histogram equalization method, and the specific implementation process is as follows: s221, dividing the denoised optical image into image blocks Sub-regions of equal size, wherein The number of blocks in the horizontal direction is the number of blocks in the horizontal direction, The number of blocks in the vertical direction; S222, local histogram statistics, namely, counting gray level histograms of the sub-areas, and calculating the pixel quantity of each gray level; S223, gray mapping transformation, namely transforming the gray value of the pixel in each subarea, wherein the transformation formula is as follows: ; Wherein, the A gray value for an output pixel; the total gray level of the image is 256 for an 8-bit image; Representation pair From 0 to Summing; Is the gray level in the subarea The number (number) of pixels of (a); is the total number (number) of pixels within the sub-region; for inputting the gray level of the pixel, the value ranges from 0 to ; S224, bilinear interpolation, namely smoothing pixels at boundaries of adjacent subareas by adopting a bilinear interpolation method, eliminating discontinuity of block boundaries, and generating an enhanced optical image.
10. The detection and recognition method based on photo-magnetic fusion according to claim 4, wherein in the step S33, the magnetic anomaly data is targeted by using an analytic signal method, and the specific implementation process is as follows: S331, calculating magnetic anomaly gradients, and respectively calculating magnetic anomaly data in Direction(s), Direction and direction Obtaining magnetic anomaly three-component gradient data by the partial derivative of the direction; S332, calculating the analytic signal amplitude, namely calculating the analytic signal amplitude of each point according to the three-component gradient data, wherein the calculation formula is as follows: ; Wherein, the Is taken as a point The resolved signal amplitude (nT/m) at; Representing an open square root operation; Is magnetic anomaly At the position of Partial derivative of direction (nT/m); Is magnetic anomaly At the position of Partial derivative of direction (nT/m); Is magnetic anomaly At the position of Partial derivative of direction (nT/m); Is the total field strength (nT) of the magnetic anomaly; 、、 the east, north and vertical coordinates (m) respectively; S333, extracting extreme points, namely searching local maximum points in the analysis signal amplitude diagram, wherein the horizontal position of the maximum points is the horizontal position of a magnetic anomaly target; s334, estimating depth, namely estimating the buried depth of the magnetic anomaly target according to the half-amplitude width of the amplitude of the analytic signal, and completing the three-dimensional positioning of the magnetic anomaly target.

Description

Detection and identification method based on photo-magnetic fusion Technical Field The invention belongs to the technical field of target detection and identification, and particularly relates to a detection and identification method based on photo-magnetic fusion. Background In the field of target detection and identification, a magnetic anomaly detection technology and an optical imaging technology are two kinds of technical means with important application values, but certain limitations exist in the application of a single means, and high-sensitivity detection and complex environment adaptability are difficult to be considered. The magnetic anomaly detection technology realizes magnetic anomaly target detection by measuring local geomagnetic field distortion caused by targets, has the advantages of all weather and medium crossing (non-magnetic medium: soil, forest, underwater and the like), but has the problems that the detection distance is limited and is easily influenced by background magnetic interference and is difficult to apply in complex environments, generally, when the detection distance is 2-3 times greater than the target size, the magnetic anomalies of the targets can be equivalent to magnetic dipoles, magnetic anomaly signals generated by the targets are attenuated along with the third power of the distance, so that deep target signals are easily submerged by shallow target signals, the magnetic anomaly forms of different types of targets on the same height surface are similar, and the like. These problems restrict the application and popularization of the magnetic anomaly detection technology in complex environments. Optical imaging techniques (e.g., visible, infrared, or laser imaging) achieve high resolution visualization by capturing reflected or radiated light signals of a target, are suitable for surface feature recognition, but are susceptible to lighting conditions, weather (e.g., fog, rain, etc.), and obstructions, and are not capable of detecting non-visible or concealed targets (e.g., underground or underwater objects). In summary, the conventional single technical means are difficult to meet the application requirements of target detection and recognition in a complex environment, and a fusion technical scheme is needed to solve the problems of sensitivity, resolution, interference resistance and the like of target detection and recognition in the complex environment. Disclosure of Invention In order to solve the problems in the background art, the invention provides a detection and identification method based on photo-magnetic fusion, which comprises the following steps: S1, acquiring data, namely carrying optical imaging equipment and magnetic measurement load by using a motion platform, acquiring optical image data and magnetic measurement data according to a planning and measuring line in a task area, wherein the magnetic measurement data comprises measurement position data and measurement height data; S2, optical data processing, namely performing image enhancement processing on the optical image data to generate an enhanced optical image, extracting surface targets in the enhanced optical image by using a target detection algorithm, and marking the type and position information of each surface target to generate an optical target detection result; S3, processing magnetic anomaly data, namely performing background magnetic field elimination and noise suppression processing on the magnetic measurement data, extracting the magnetic anomaly data, then acquiring position and contour information of a magnetic anomaly target by utilizing a target positioning algorithm, and then converting the magnetic anomaly data into regular grid data through grid interpolation processing to generate a magnetic anomaly distribution map; S4, photo-magnetic data fusion processing, namely superposing the magnetic anomaly distribution map on the enhanced optical image, performing space matching on the optical target detection result and the position of the magnetic anomaly target, performing preliminary classification on the target according to the matching result, performing magnetic field modeling on the visible ferromagnetic target with larger magnetic anomaly intensity, calculating magnetic anomaly distribution generated in a task area, subtracting the magnetic anomaly distribution generated by the visible ferromagnetic target from total magnetic anomaly data of the task area to obtain residual magnetic anomaly data, and enhancing magnetic anomaly signals of a hidden target; S5, target identification and output, namely inputting the data subjected to the photomagnetic fusion into a classification model for target identification, and outputting the target type, the target position profile and the magnetic characteristic parameters. Further, the step S1 includes the following substeps: S11, line planning, namely setting line height and line interval parameters according to a task area range, fli