CN-121640281-B - Ecological protection area-oriented intelligent identification method for invasive unmanned aerial vehicle

CN121640281BCN 121640281 BCN121640281 BCN 121640281BCN-121640281-B

Abstract

The invention relates to the technical field of image recognition, in particular to an intelligent recognition method of an invading unmanned aerial vehicle facing an ecological protection area, which is used for judging whether an acoustic detection device is triggered at each moment according to the static characteristic and the dynamic characteristic of a heat source target in an infrared image at each moment of the ecological protection area, processing a time domain sound signal into a Mel frequency spectrum chart after the acoustic detection device is triggered, determining the energy characteristic of the Mel frequency spectrum chart according to the uniform condition of energy distribution in each Mel frequency band, and determining the invasion recognition result of the unmanned aerial vehicle according to the change association condition of the energy characteristic of the heat source target and the running speed. According to the invention, the primary screening trigger control is carried out through the infrared image, then the acoustic fine analysis is carried out, and finally the unmanned aerial vehicle intrusion recognition result is obtained according to the cross-modal physical association fusion, so that the recognition precision is improved, the false alarm rate is reduced, the energy efficiency is improved, the complex acoustic and thermal background interference in the ecological protection area can be effectively treated, and the environmental adaptability is improved.

Inventors

HONG JUNCHENG
XU JIANWEI
ZHANG YUANJING
ZHANG JUNSHEN

Assignees

浙大启真未来城市科技(杭州)有限公司
浙江大学城乡规划设计研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20260202

Claims (9)

1. An intelligent identification method of an intrusion unmanned aerial vehicle facing an ecological protection area is characterized by comprising the following steps: Judging whether the acoustic detection equipment is triggered at each moment according to the static characteristics and the dynamic characteristics of the heat source target in the infrared image at each moment of the ecological protection area, wherein the static characteristics comprise shapes and temperature distribution, and the dynamic characteristics are motion tracks; after triggering the acoustic detection equipment, processing the time domain sound signals at all moments acquired by the acoustic detection equipment into a Mel spectrogram; determining the energy characteristics of the Mel spectrogram according to the uniformity of energy distribution in each Mel frequency band in the Mel spectrogram; according to the change association condition of the energy characteristics and the running speed of the heat source target, the invasion possibility of the unmanned aerial vehicle in the ecological protection area is obtained; determining an unmanned aerial vehicle intrusion recognition result according to the unmanned aerial vehicle intrusion possibility; according to the change association condition of the energy characteristics and the running speed of the heat source target, the unmanned aerial vehicle invasion possibility of the ecological protection area is obtained, and the unmanned aerial vehicle invasion possibility comprises the following steps: determining the energy characteristic difference of the mel spectrograms at two adjacent moments and the running speed difference of the heat source targets at two adjacent moments; and obtaining the invasion possibility of the unmanned aerial vehicle in the ecological protection area aiming at the adjacent two moments according to the difference between the energy characteristic difference and the running speed difference, wherein the invasion possibility of the unmanned aerial vehicle is inversely related to the difference between the energy characteristic difference and the running speed difference.
2. The intelligent recognition method of an intrusion unmanned aerial vehicle facing an ecological protection area according to claim 1, wherein the judging whether the acoustic detection device is triggered at each moment according to the static characteristic and the dynamic characteristic of the heat source target in the infrared image at each moment of the ecological protection area comprises: Obtaining the shape regularity of the heat source target at each moment according to the shape of the heat source target in the infrared image at each moment; determining the temperature distribution confusion of a heat source target in an infrared image at each moment, and combining the shape regularity to obtain the unmanned aerial vehicle suspected degree of the heat source target at each moment; Obtaining the motion track confusion degree of each moment of the heat source target based on the motion curvature difference of two adjacent moments in the motion track of the local time window of each moment of the heat source target; according to the suspected degree of the unmanned aerial vehicle and the chaotic degree of the motion track, obtaining the acoustic detection triggering possibility of the heat source target at each moment, wherein the acoustic detection triggering possibility is positively correlated with the suspected degree of the unmanned aerial vehicle and the chaotic degree of the motion track; And judging whether the acoustic detection equipment is triggered at each moment according to the acoustic detection triggering possibility of the heat source target at each moment.
3. The intelligent recognition method for an intrusion unmanned aerial vehicle facing an ecological protection area according to claim 2, wherein the process of obtaining the shape regularity comprises the following steps: determining a minimum circumscribed rectangle of a heat source target in the infrared image; And acquiring the width-to-length ratio of the minimum circumscribed rectangle as the shape regularity of the heat source target.
4. The intelligent recognition method for an intrusion unmanned aerial vehicle facing an ecological protection area according to claim 2, wherein the acquiring process of the temperature distribution confusion comprises the following steps: Determining the temperature gradient characteristics of each pixel point of the heat source target in the infrared image; Obtaining the influence weight of the temperature gradient characteristic of each pixel point according to the temperature value of each pixel point of the heat source target in the infrared image, wherein the influence weight is positively correlated with the temperature value; and carrying out weighted summation on the temperature gradient characteristics of each pixel point according to the influence weights of the temperature gradient characteristics of each pixel point to obtain the temperature distribution confusion of the heat source target in the infrared image at each moment.
5. The intelligent recognition method for an intrusion unmanned aerial vehicle facing an ecological protection area according to claim 2, wherein the process for obtaining the degree of confusion of the motion trail comprises the following steps: Obtaining the moving curvature difference characteristics of two adjacent moments according to the moving curvature difference of the two adjacent moments in the local time window of each moment of the heat source target; Determining the number of moving pixel points in the infrared image of the heat source target at each moment based on a frame difference method; Obtaining reference weights of moving curvature difference characteristics of two adjacent moments in the local time window of each moment of the heat source target according to the quantity difference of the moving pixel points of the two adjacent moments in the local time window of each moment of the heat source target; And according to the reference weights of the moving curvature difference characteristics of the adjacent two moments in the local time window of each moment of the heat source target, carrying out weighted summation on the moving curvature difference characteristics of the adjacent two moments in the local time window of each moment of the heat source target, and obtaining the motion track confusion degree of each moment of the heat source target.
6. The intelligent recognition method of an intrusion unmanned aerial vehicle facing an ecological protection area according to claim 2, wherein the judging whether the acoustic detection device is triggered at each moment according to the acoustic detection triggering possibility of the heat source target at each moment comprises determining to trigger the acoustic detection device when the acoustic detection triggering possibility is larger than a preset triggering possibility threshold.
7. The intelligent recognition method of the intrusion unmanned aerial vehicle facing the ecological protection area according to claim 1, wherein the processing of the time domain sound signals of all the moments acquired by the acoustic detection equipment into the Mel spectrogram comprises the processing of the time domain sound signals of the local time window of all the moments into the Mel spectrogram of all the moments.
8. The intelligent recognition method for an intrusion unmanned aerial vehicle facing an ecological protection area according to claim 1, wherein the acquiring process of the energy characteristics of the mel spectrogram comprises the following steps: graying each mel frequency band in the mel frequency spectrum map to obtain a regional gray map of each mel frequency band; obtaining the energy distribution uniformity degree of each Mel frequency band according to the gray value standard deviation of the regional gray map of each Mel frequency band; And when the maximum energy distribution uniformity degree meets a preset condition, acquiring energy of a Mel frequency band corresponding to the maximum energy distribution uniformity degree, and obtaining energy characteristics of the Mel spectrogram.
9. The intelligent recognition method of an intrusion unmanned aerial vehicle for an ecological protection area according to claim 1, wherein the step of determining the unmanned aerial vehicle intrusion recognition result according to the unmanned aerial vehicle intrusion possibility comprises the step of judging that the unmanned aerial vehicle intrusion condition exists in the ecological protection area when the continuous preset number of unmanned aerial vehicle intrusion possibilities are all larger than a preset intrusion possibility threshold value.

Description

Ecological protection area-oriented intelligent identification method for invasive unmanned aerial vehicle Technical Field The invention relates to the technical field of image recognition, in particular to an intelligent recognition method of an invasive unmanned aerial vehicle facing an ecological protection area. Background When the unmanned aerial vehicle that invades ecological protection district is discerned, current recognition mode is the mode of acoustic detection generally, sets up microphone array in a plurality of positions in ecological protection district promptly, gathers the sound signal that unmanned aerial vehicle screw produced to detect unmanned aerial vehicle invasion condition. However, acoustic detection is susceptible to environmental noise, such as wind noise, animal sounds, etc., resulting in lower recognition accuracy. In order to prevent the unmanned aerial vehicle from interfering with the ecological protection area, effective identification needs to be carried out on the invasive unmanned aerial vehicle. The prior art mainly comprises: the infrared image-based recognition method has the advantages that the thermal imaging camera is used for capturing the heat radiation of the unmanned aerial vehicle engine or the propeller, and the defects that a large number of moving heat sources such as wild animals exist in an ecological protection area, misjudgment is easy to generate by simply relying on visual characteristics, and the resolution capability of the heat source of a remote or small unmanned aerial vehicle is limited. The characteristic voiceprint of the unmanned aerial vehicle propeller is collected through the microphone array based on the identification of acoustic detection, and the defect is that background noise such as wind noise, animal sound, water flow sound and the like in natural environment can seriously interfere with or even submerge a target signal, so that the identification rate is reduced. Any of the above single techniques is difficult to achieve stable and accurate identification in complex scenarios such as ecological protection zones. Although a scheme for simply and parallelly using a plurality of sensors exists, the system can continuously run with high power consumption, the data processing burden is heavy, and the problems of effective coordination and decision fusion among different modality information cannot be solved. Disclosure of Invention The invention aims to solve the technical problem of providing the intelligent identification method of the invasive unmanned aerial vehicle, which can adapt to the complex environment of an ecological protection area, effectively integrate infrared and acoustic information, ensure high identification accuracy and reduce the energy consumption and false alarm rate of the system. The invention aims to provide an intelligent identification method of an invasion unmanned aerial vehicle facing an ecological protection area, which adopts the following technical scheme: The invention provides an intelligent identification method of an intrusion unmanned aerial vehicle facing an ecological protection area, which comprises the following steps: Judging whether the acoustic detection equipment is triggered at each moment according to the static characteristics and the dynamic characteristics of the heat source target in the infrared image at each moment of the ecological protection area, wherein the static characteristics comprise shapes and temperature distribution, and the dynamic characteristics are motion tracks; after triggering the acoustic detection equipment, processing the time domain sound signals at all moments acquired by the acoustic detection equipment into a Mel spectrogram; determining the energy characteristics of the Mel spectrogram according to the uniformity of energy distribution in each Mel frequency band in the Mel spectrogram; according to the change association condition of the energy characteristics and the running speed of the heat source target, the invasion possibility of the unmanned aerial vehicle in the ecological protection area is obtained; And determining an unmanned aerial vehicle intrusion recognition result according to the unmanned aerial vehicle intrusion possibility. In an exemplary embodiment, the determining whether each moment triggers the acoustic detection device according to the static feature and the dynamic feature of the heat source target in the infrared image of each moment in the ecological protection area includes: Obtaining the shape regularity of the heat source target at each moment according to the shape of the heat source target in the infrared image at each moment; determining the temperature distribution confusion of a heat source target in an infrared image at each moment, and combining the shape regularity to obtain the unmanned aerial vehicle suspected degree of the heat source target at each moment; Obtaining the motion track confusion degree of each m