A wideband specific frequency point signal spectrum resource detection method based on image processing

Abstract

The present application relates to a kind of broadband specific frequency point signal spectrum resource detection methods based on image processing, comprising: the data collected is converted into spectral image by FFT, simultaneously, the collected data is cached processing.Through the analysis and processing of spectral image, the effective data in specific frequency range is determined, signal detection result set is obtained, if effective data is detected, then alarm monitoring is carried out, simultaneously, the cached data is stored, otherwise, the cached data is cleared.The present application converts IQ data analysis and processing into image processing, and according to the image pixel distribution characteristics, the image noise floor data characteristics are statistically modeled, the precision and processing speed of image noise floor estimation are greatly improved, and the specific signal monitoring task with certain anti-interference ability is realized.At the same time, the limitation of high performance requirement for non-cooperative communication equipment is broken, and in the case of relatively poor electromagnetic environment, good monitoring effect can still be maintained, and the collection of effective width data is realized.

Description

Technical Field

[0001] This invention relates to the field of signal recognition technology, and in particular to a method for detecting broadband specific frequency signal spectrum resources based on image processing. Background Technology

[0002] With the continuous development of communication systems, broadband signals have become an important component. In actual communication environments, factors such as weather, interference, and attenuation often make the electromagnetic environment more complex. These signals may be affected by noise and interference during transmission, causing changes in the signal's spectral composition, affecting signal clarity and reliability, and thus increasing the difficulty of non-cooperative communication tasks. A primary task in non-cooperative communication is to detect narrowband signals at key frequency points of the broadband channel, achieving narrowband signal monitoring. Therefore, accurately detecting the spectral components and signal characteristics of broadband signals is crucial for improving the performance of non-cooperative communication systems and communication monitoring. In many application scenarios, maintaining non-cooperative communication capabilities requires not only more communication equipment but also avoiding the negative impacts of complex electromagnetic environments on broadband signal transmission. This often places higher demands on the functionality and performance of data acquisition and processing equipment, thereby increasing costs. Simultaneously, complex electromagnetic environments undoubtedly increase the difficulty of broadband signal monitoring in non-cooperative communication tasks.

[0003] At present, the mainstream broadband detection methods are mainly based on the processing methods of the original signal level characteristics and signal features, such as wavelet domain detection and bandpass filtering. Although they can achieve certain results, they are too dependent on the performance of hardware devices, and have problems such as poor monitoring effect and slow processing speed in non-cooperative communication broadband signals in complex electromagnetic environments.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a broadband specific frequency signal spectrum resource detection method based on image processing, which solves the deficiencies of the existing methods.

[0006] The objective of this invention is achieved through the following technical solution: a method for detecting the spectrum resources of broadband signals at specific frequency points based on image processing, the monitoring method comprising:

[0007] The acquired IQ data is converted into spectral image data using FFT, and the corresponding IQ data is cached.

[0008] A first frequency range is defined, and the effective spectrum image data within the first frequency range is analyzed and processed. Image texture features are extracted from the effective spectrum image data to obtain a signal texture feature set.

[0009] The signal texture feature set is judged according to the effective signal texture discrimination criteria to obtain the texture set to be classified, and the texture set to be classified is processed to obtain the candidate signal set;

[0010] Set a second frequency range and filter the selected signal set within the second frequency range to obtain the signal detection results under the current spectrum data. If valid data is detected, alarm monitoring is performed, and the cached data is written to disk; otherwise, the cached data is cleared.

[0011] The process of setting a first frequency range, analyzing and processing the effective spectrum image data within that range, and extracting image texture features from the effective spectrum image data to obtain a signal texture feature set specifically includes the following:

[0012] Set a first frequency range, and determine the frequency range R of each specified frequency point with the first frequency range as the center and the first frequency range as the radius. Based on the distribution characteristics of pixels, perform statistical modeling on the image noise data characteristics, use local noise estimation to obtain the local noise value, and solve the noise change trend in the frequency range R of each frequency point in the frequency point library in turn based on the noise value and curve interpolation to obtain the noise function.

[0013] The pixel data within the frequency range R of each specified frequency point are statistically compared according to the noise floor function. Pixel information below the noise floor setting value is retrieved according to the set filtering criteria. The retrieved noise floor within the frequency range R of each specified frequency point is filtered to extract the image texture features.

[0014] The extracted image texture features are subjected to continuity analysis, and texture features with close pixel distances are filled and merged to obtain the signal texture feature set U.

[0015] The process of determining the texture set to be classified based on the effective signal texture discrimination criteria includes: filtering effective signals from the signal texture feature set U based on two effective signal texture discrimination criteria, namely the minimum duration of the signal and the minimum representation intensity of the signal, to obtain the candidate signal texture set U'.

[0016] The process of processing the texture set to be classified to obtain the candidate signal set includes:

[0017] Based on the position and size of the texture, each segment of texture in the candidate signal texture set U' is classified to obtain the candidate classification texture set U;

[0018] The image locations of various signal textures in the texture set U'' to be classified constitute class boundaries. Based on the pixel distance and positional relationship between class boundaries, intra-class merging of various signal textures is performed to obtain the candidate signal set V.

[0019] The process involves statistically modeling the image noise data features, using local noise estimation to obtain local noise values, and then sequentially solving for the noise variation trend within the frequency range R of each frequency point in the frequency point library based on the noise values ​​and curve interpolation. The resulting noise function includes:

[0020] Based on the frequency point library P, within a specified range of frequency point pi, a sliding window with a local window length of k is set. Within the frequency range R, a sliding window operation with a step size of s is performed on each column of the image matrix. The maximum value m is obtained during each sliding window operation. i Find the n maximum values ​​in each column to obtain set M;

[0021] Find the mode c of set M i As a representative of the noise floor, r noise floor sets X are obtained within the frequency range R specified by the frequency point pi. Spline interpolation is performed on the noise floor sets X to obtain the trend function f of the noise floor.

[0022] The present invention has the following advantages:

[0023] 1. By converting IQ data into a spectrogram, the traditional signal level analysis problem is transformed into an image processing problem, reducing the complexity and difficulty of data analysis and processing. It can also achieve the purpose of effective signal detection and data monitoring for broadband data with low signal-to-noise ratio, and has a certain anti-interference capability. It improves the flexibility and adaptability of signal monitoring in non-cooperative communication tasks in complex electromagnetic environments.

[0024] 2. By converting IQ data analysis and processing into image processing, and statistically modeling the image noise data characteristics based on image pixel distribution features, the accuracy and processing speed of image noise estimation are greatly improved, enabling specific signal monitoring tasks with a certain degree of anti-interference capability. Simultaneously, it breaks the limitation of high-performance requirements for non-cooperative communication equipment, maintaining good monitoring performance even in harsh electromagnetic environments, and achieving the acquisition of effective bandwidth data. Compared to traditional technologies, the process for broadband data monitoring and acquisition is more rational, and the technology is more practical. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the overall process of the present invention;

[0026] Figure 2 This is a schematic diagram of the spectrum image analysis and processing flow in this invention;

[0027] Figure 3This is the processing flow of the spectral image feature modeling and noise floor estimation algorithm in this invention;

[0028] Figure 4 This is a schematic diagram of the signal texture continuity processing flow in this invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of this application provided below with reference to the accompanying drawings is not intended to limit the scope of protection of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. The present invention will be further described below with reference to the accompanying drawings.

[0030] like Figure 1 As shown, this invention specifically relates to a method for detecting broadband signal spectrum resources at specific frequency points based on image processing. The method converts acquired data into a spectrum image using FFT and simultaneously caches the acquired data. Through analysis and processing of the spectrum image, valid data within a specific frequency range is determined, resulting in a signal detection result set. If valid data is detected, an alarm is triggered, and the cached data is stored on disk; otherwise, the cached data is cleared. Specifically, it includes the following:

[0031] Step 1: First, convert the accumulated IQ data into spectrum image data of a certain specification using FFT (Fast Fourier Transform), and then cache the corresponding IQ data.

[0032] Step 2: Set a specific frequency range R with a wide location range (e.g., within a specified frequency range of 5M). Based on the distribution characteristics of pixels, perform statistical modeling on the image noise data characteristics. Use local noise estimation to obtain the local noise value. Based on the noise value and curve interpolation, solve the noise change trend in each frequency range R within the specific frequency library in turn to obtain the noise function.

[0033] Step 3: Statistically compare the pixel data in the R domain of each frequency point according to the noise floor function, retrieve pixel information that is lower than the noise floor set value according to the set filtering criteria, filter the retrieved noise floor of each frequency point in the R domain, and realize the extraction of image texture features.

[0034] Step 4: Perform continuity analysis on the extracted image texture features, fill and merge texture features with close pixel distances, and obtain the signal texture feature set U;

[0035] Step 5: Based on the custom valid signal texture discrimination criteria ① minimum signal duration and ② minimum signal representation intensity, filter valid signals from the signal texture feature set to obtain the candidate signal texture set U;

[0036] Step 6: Based on the position and size of the textures, classify each segment of the texture in the candidate signal texture set U to obtain the candidate classification texture set U;

[0037] Step 7: Determine the image positions of various signal textures in the candidate classification texture set to form class boundaries, and merge the intra-class signal textures according to the pixel distance and positional relationship between class boundaries to obtain the candidate signal set V;

[0038] Step 8: Set a more stringent location range (e.g., within a specified frequency range of 1MHz) for a specific frequency range R. c The signal detection results under the current spectrum data are obtained by further filtering the candidate signal set V. If no signal is detected in the results, proceed to step nine; if a signal is detected, proceed to step ten.

[0039] Step 9: Clear cached data;

[0040] Step 10: Store the current cached data on disk, mark the frequency points where the signals exist, and issue an alarm to achieve signal monitoring at specific frequency points of the current spectrum data.

[0041] like Figure 2 As shown, statistical modeling of pixel distribution in a spectral image within a specific frequency range is performed, and the noise floor function within the range R of each frequency point in the specific frequency point library is solved sequentially, such as... Figure 3 As shown, based on the frequency point library P, a sliding window with a local window length of k is set within a specified range of frequency point pi. Within the frequency range R, a sliding window operation with a step size of s is performed on each column of the image matrix. The maximum value m is obtained during each sliding window operation. i Finding the n maximum values ​​in each column yields a set M of {m1, m2, m3, ..., m}. n Find the mode c of set M. i As a representative of the noise floor, r noise floor sets X are obtained within the frequency range R specified by the frequency point pi. Spline interpolation is performed on the noise floor sets X to obtain the trend function f of the noise floor.

[0042] Based on the noise floor function, noise filtering and feature extraction are performed on the pixel data in the R domain at each frequency point, such as... Figure 4As shown, continuity analysis is performed on the extracted image texture features to obtain a signal texture feature set U. Based on the custom effective signal texture discrimination criteria, including the minimum signal duration and the minimum signal representation intensity, effective signals are filtered from the signal texture feature set to obtain a candidate signal texture set. According to the position and size of the texture, each segment of texture in the candidate signal texture set is classified to obtain a candidate classification texture set. The class boundaries of each type of signal texture in the candidate classification texture set are solved, and intra-class merging of each type of signal texture is performed according to the class boundaries to obtain a signal detection result set.

[0043] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and improvements, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A method for detecting wideband specific frequency point signal spectrum resource based on image processing, characterized in that: The detection method includes: The acquired IQ data is converted into spectral image data using FFT, and the corresponding IQ data is cached. A first frequency range is defined, and the effective spectrum image data within the first frequency range is analyzed and processed. Image texture features are extracted from the effective spectrum image data to obtain a signal texture feature set. The signal texture feature set is judged according to the effective signal texture discrimination criteria to obtain the texture set to be classified, and the texture set to be classified is processed to obtain the candidate signal set; Set a second frequency range and filter the selected signal set within the second frequency range to obtain the signal detection results under the current spectrum data. If valid data is detected, an alarm is triggered and the cached data is written to disk; otherwise, the cached data is cleared. The process of setting a first frequency range, analyzing and processing the effective spectrum image data within that range, and extracting image texture features from the effective spectrum image data to obtain a signal texture feature set specifically includes the following: Set a first frequency range, take each specified frequency point as the center and the first frequency range as the radius, determine the frequency range domain R of each specified frequency point, perform statistical modeling on the image noise data features according to the pixel distribution characteristics, use local noise estimation to obtain the local noise value, and solve the noise change trend in the frequency range domain R of each frequency point in the frequency point library according to the noise value and curve interpolation to obtain the noise function. The pixel data within the frequency range R of each specified frequency point are statistically compared according to the noise floor function. Pixel information below the noise floor setting value is retrieved according to the set filtering criteria. The retrieved noise floor within the frequency range R of each specified frequency point is filtered to extract the image texture features. The extracted image texture features are subjected to continuity analysis, and texture features with close pixel distances are filled and merged to obtain the signal texture feature set U; The process involves statistically modeling the image noise data features, using local noise estimation to obtain local noise values, and then sequentially solving for the noise variation trend within the frequency range R of each frequency point in the frequency point library based on the noise values ​​and curve interpolation. The resulting noise function includes: Based on the frequency point library P, within a specified range of frequency point pi, a sliding window with a local window length of k is set. Within the frequency range R, a sliding window operation with a step size of s is performed on each column of the image matrix, and the maximum value is obtained during each sliding window operation. Find the n maximum values ​​in each column to obtain set M; Find the mode of set M As a representative of the noise floor, r noise floor sets X are obtained within the frequency range R specified by the frequency point pi. Spline interpolation is performed on the noise floor sets X to obtain the trend function f of the noise floor.

2. The method for detecting broadband specific frequency signal spectrum resources based on image processing according to claim 1, characterized in that: The process of obtaining the texture set to be classified by judging the signal texture feature set according to the effective signal texture discrimination criteria includes: filtering effective signals from the signal texture feature set U according to the two effective signal texture discrimination criteria of minimum signal duration and minimum signal representation intensity to obtain the candidate signal texture set U'.

3. The method for detecting broadband specific frequency signal spectrum resources based on image processing according to claim 2, characterized in that: The process of processing the texture set to be classified to obtain the candidate signal set includes: Based on the position and size of the texture, each segment of texture in the candidate signal texture set U' is classified to obtain the candidate classification texture set U''. The image locations of various signal textures in the texture set U'' to be classified constitute class boundaries. Based on the pixel distance and positional relationship between class boundaries, intra-class merging of various signal textures is performed to obtain the candidate signal set V.

Images