37results about How to "Improve target detection probability" patented technology

The invention discloses a distributed passive radar target detection realization method, and the method belongs to the technical field of passive radar target detection. An existing target detection method is characterized by carrying out correlation and over-threshold processing independently on each bistatic pair to finish target detection, and then, obtaining bistatic distance and Doppler information of a target through estimation and carrying out target location, which results in the problem of complex ambiguous location needing to be solved in the follow-up target exact positioning process. In order to solve the problem above, the invention introduces target location and velocity vector and constructs centralized target detection statistics, so that target positioning is realized while realizing target detection, and the problem of ambiguous target location no longer needs to be treated. Besides, the method carries out non-phase-coherent accumulation on a plurality of transmitter-receiver pairs before detection to obtain space diversity gains, thereby improving target detection probability, obtaining more-stable target detection performance and providing basis for continuous and stable tracking of the target.

The invention discloses a multi-target detection and positioning method based on building perspective radar imaging. The method includes back projection imaging, superimposed binarized images, target enhancement, acquisition of image matrix accumulation vectors, target position extraction, and target area normalization. . The invention uses the backward projection imaging method to speed up the imaging of the area to be detected, and realizes the target correlation between consecutive frames after the imaging of the non-connected domain detection by superimposing the binarized image and the target enhancement processing, and enhances the target amplitude, which is beneficial to the subsequent detection. The multi-target enhancement detection based on image directionality is realized by acquiring image matrix accumulation vector, target matching extraction, and target area normalization processing, which effectively enhances targets with low echo amplitudes covered by strong targets, thereby improving weaker targets. The detection rate of the target has the advantages of good real-time performance and high target detection probability, and has extremely high application value in the field of building perspective radar technology.

The invention discloses a frequency diversity array radar low-altitude target detection method, which belongs to the field of phased array radar low-altitude target detection. First of all, the scanning characteristics of the FDA radar make its beam pointing related to the radial distance and the frequency difference. By introducing a frequency difference between different transmitting array elements, the signal-to-noise ratio caused by the multipath effect in the low-altitude target environment can be effectively suppressed. loss. Then, the generalized likelihood ratio test (GLRT) method is used to derive the FDA radar low-altitude target detector, which can effectively improve the target detection probability under the condition of a certain false alarm rate. Finally, through the design of sub-modules such as matrix multiplication and matrix determinant calculation, the FDA radar low-altitude target detector is realized on the FPGA hardware platform, which effectively improves the real-time performance of FDA radar low-altitude target detection. To sum up, this method can effectively complete the detection of low-altitude targets by using the scanning characteristics of FDA radar in the low-altitude target environment, and has high practicability in modern warfare.

The invention discloses a multi-channel time-frequency accumulation processing method for instantaneous broadband multi-frequency composite radar signals, which includes: target scattered echoes are sent to a band-pass filter group to filter out each of the transmitted multi-frequency signals after passing through a low-noise radio frequency receiving link Carrier modulation spectrum; each carrier modulation spectrum is down-converted to the baseband with the corresponding carrier signal, and the pulse pressure processing is performed through a module matching the constant envelope modulation signal; each frequency channel is sent to the Doppler correction processing module after pulse pressure, Solve the problem of echo crossing Doppler units on different carrier frequency channels; the corrected channel signals are sent to the logic selection synthesis module to identify the interference of each frequency channel, select the undisturbed channel, and use non-coherent synthesis The method completes the accumulation of echo energy on the range-Doppler two-dimensional map. The invention can solve the Doppler frequency shift and the corresponding inter-unit problem, and improve the anti-jamming and anti-interception performance of the radar.

The invention discloses a method for detecting weak spot targets in an infrared image, which adopts structural elements of 5×5 dimensions in 8 directions to extract all possible targets of different scales; obtains the target region of interest through adaptive threshold processing, and effectively improves the target Signal-to-noise ratio; use the judgment criterion of background edge points and point targets in the local neighborhood to remove the residual background edge points; remove noise points and obtain the target motion trajectory according to the matching relationship between frames; the algorithm complexity is not high and meets real-time The results show that the target detection probability of the present invention is high, and the false alarm rate is low at the same time, which has important application value.

The invention discloses a detection method for radar target long-time accumulation based on a linear frequency modulation continuous wave. The method comprises the following steps: firstly, confirming a to-be-searched space according to a motion model of a to-be-searched target; emitting a linear frequency modulation continuous wave signal, and performing dechirping treatment on a received radar target echo, thereby acquiring the target echo after dechirping; performing FFT (Fast Fourier Transform) on the target echo after dechirping along a fast time dimension, thereby acquiring an echo signal of a slow time-fast time frequency domain; for the echo signal, performing phase-coherent accumulation on each motor parameter vector in the searching space, thereby acquiring evaluation values; and utilizing the evaluation values to judge a threshold. According to the method disclosed by the invention, through long-time phase parameter accumulation, the signal-to-noise ratio after the target accumulation is effectively increased under the condition of limited emitter power, so that the detection performance of the target can be promoted.

The invention discloses an infrared target detection method, device, equipment and medium based on multi-feature fusion. The method includes acquiring a detection target image; performing global threshold segmentation on the detection target image to obtain a binarized image; Connected domain marking is performed to obtain candidate targets; feature extraction is performed on each candidate target, and each feature channel is filtered to obtain the feature component value of each candidate target; multi-feature fusion is performed on candidate targets to obtain a normalized feature vector ; Judging whether the normalized feature vector reaches the detection threshold, if it reaches the detection threshold, the candidate object is retained, otherwise, the candidate object is deleted. Aiming at the problems of low contrast of infrared images and less target feature information, the present invention performs multi-feature fusion processing by mining various feature information of targets and clutter, and utilizes multiple feature channels for parallel processing, thereby improving target detection probability and reducing false alarm rate .

The present invention proposes a special-shaped radar joint judgment target detection method, respectively adopts the high and low thresholds to perform CFAR detection on the original video data of each radar, and extracts the distance cohesion point trace from the data after the high and low threshold detection of each radar, and according to the local information of the distance point trace The clutter / noise ratio and the noise floor of each radar are used to normalize the trace amplitude, and the joint detection and judgment of the low-threshold detection condensation point traces of all radars are carried out according to the radar distance measurement error, and the high-threshold detection condensation point traces of multiple radars are detected. Combined to complete the original video-level multi-radar joint target detection processing. The present invention adopts two kinds of thresholds, high and low, which improves the probability of target detection. For each radar, the distance point trace is first extracted and then jointly judged, which avoids the failure of direct fusion processing caused by the difference in multi-radar sampling rate, resolution, noise floor, etc. question.