Insect sound detecting system and method thereof

Abstract

An insect sound detecting system is proposed. The insect sound detecting system includes a sound receiving device, a storing device and a processing device. The sound receiving device is configured to receive a voice from an object to generate a sound signal. The storing device stores the sound signal. The processing device receives the sound signal, and includes a feature extracting module and an artificial intelligent module. The feature extracting module extracts a plurality of feature information from the sound signal. The artificial intelligent module is signally connected to the feature extracting module, inputs the feature information into a first detecting model to generate a first detecting result, and inputs the feature information into a second detecting model to generate a second detecting result. The first detecting result includes whether there is an insect in the object. The second detecting result includes a species of the insect. Thus, the insect sound detecting system of the present disclosure can monitor there is any insect on wood products.

Description

[Technical Field]

[0001] This invention relates to a sound wave detection system and method, and more particularly to a sound wave detection system and method for insect infestation. [Previous Technology]

[0002] When wooden objects or trees are infested by wood-boring insects, apart from the visible parts that can be identified through manual inspection, the internal damage caused by wood-boring insects cannot be judged or confirmed by the naked eye. If a tree or wooden object is infested by wood-boring insects, its internal structure may become fragile or break, leading to the death of the tree or the destruction of the wooden object.

[0003] Existing methods for detecting woodworms include destructive and non-destructive methods. Destructive methods include detection through tapping, wood moisture content meters, infrared thermal imagers, X-ray scanning, resistivity measurement, magnetic resonance imaging, and radioactive isotope scanning. However, these methods require specialized instruments and equipment, and their costs are high under long-term testing conditions.

[0004] In view of this, there is currently a lack of a non-invasive, low-cost acoustic detection system and method for insect infestation on the market. [Summary of the Invention]

[0005] Therefore, the object of the present invention is to provide an acoustic detection system and method for insect infestation, which uses a sound receiving device to collect sound from an object and a processing device to identify whether the object contains insects and to identify the type of insect. Thus, the acoustic detection system and method of the present invention can detect whether wooden objects are infested by insects in a non-invasive manner.

[0006] According to one embodiment of the structural configuration of the present invention, an insect-boring sound wave detection system is provided, comprising a sound receiving device, a storage device, and a processing device. The sound receiving device is used to receive sound from an object to generate a sound signal. The storage device is signal-connected to the sound receiving device and stores the sound signal. The processing device is signal-connected to the storage device and receives the sound signal. The processing device includes a feature extraction module and an artificial intelligence module. The feature extraction module extracts multiple feature information from the sound signal. The artificial intelligence module is signal-connected to the feature extraction module, inputs this feature information into a first recognition model to generate a first recognition result, and inputs this feature information into a second recognition model to generate a second recognition result. The first recognition result includes whether there is an insect in the object, and the second recognition result includes a category of insect.

[0007] Other embodiments of the aforementioned implementation are as follows: The aforementioned feature extraction module may include a segmentation module, a processing module, a Fourier transform module, and a feature transformation module. The segmentation module segments the sound signal into a complex number of sound segment signals. The processing module signal is connected to the segmentation module and performs a window function processing on these sound segment signals to generate complex segment windowed signals. The Fourier transform module signal is connected to the processing module and performs a Fourier transform on these segment windowed signals to convert them from the time domain to the frequency domain. The feature transformation module signal is connected to the Fourier transform module and transforms these segment windowed signals through a Mel-scale filter to generate these feature information.

[0008] Other embodiments of the foregoing implementation are as follows: either the first identification model or the second identification model may include a Shifted Windows Transformer (Swin Transformer).

[0009] Other embodiments of the foregoing implementation are as follows: The aforementioned radio receiver may include a listening probe, a sensor encapsulating a surface acoustic wave filter, or a piezoelectric thin film electronic stethoscope.

[0010] Other embodiments of the foregoing implementation are as follows: The foregoing processing device may further include a collection module. The collection module is signal-connected to the artificial intelligence module, and marks the first identification result and the second identification result corresponding to the object on the sound signal and stores them in the storage device.

[0011] According to one embodiment of the method of the present invention, a method for detecting insect infestation using sound waves is provided, comprising: driving a sound receiving device to receive sound from an object to generate a sound signal; driving a storage device to store the sound signal; driving a feature extraction module of a processing device to extract multiple feature information from the sound signal; driving an artificial intelligence module of the processing device to input the feature information into a first recognition model to generate a first recognition result; and driving the artificial intelligence module of the processing device to input the feature information into a second recognition model to generate a second recognition result. The first recognition result includes whether there is an insect in the object, and the second recognition result includes a category of insect infestation.

[0012] Other embodiments of the foregoing implementation are as follows: The foregoing insect-boring sound wave detection method may further include a segmentation module driving the feature extraction module to segment the sound signal into a complex number of sound segment signals; a processing module driving the feature extraction module to perform a window function processing on these sound segment signals to generate complex segment windowed signals; a Fourier transform module driving the feature extraction module to perform a Fourier transform on these segment windowed signals to convert them from the time domain to the frequency domain; and a feature conversion module driving the feature extraction module to convert these segment windowed signals through a Mel-scale filter to generate these feature information.

[0013] Other embodiments of the foregoing implementation are as follows: either the first identification model or the second identification model may include a moving window converter.

[0014] Other embodiments of the foregoing implementation are as follows: The aforementioned radio receiver may include a listening probe, a sensor encapsulated with a surface acoustic wave filter, or a piezoelectric thin film electronic stethoscope.

[0015] Other embodiments of the foregoing implementation are as follows: The foregoing insect infestation sound wave detection method may further include a collection module of the driving processing device to mark the first identification result and the second identification result corresponding to the object on the sound signal and store them in the storage device.

Implementation Method

[0016] Several embodiments of the present invention will now be described with reference to the drawings. For clarity, many practical details will be set forth in the following description. However, it should be understood that these practical details are not intended to limit the invention. That is, in some embodiments of the invention, these practical details are not essential. Furthermore, for the sake of simplicity, some conventional structures and elements will be shown in the drawings in a simple schematic manner; and repeated elements may be denoted by the same number.

[0017] Furthermore, in this document, when a component (or unit or module, etc.) is "connected" to another component, it can mean that the component is directly connected to the other component, or it can mean that the component is indirectly connected to the other component, that is, there is another component between the component and the other component. Only when it is explicitly stated that a component is "directly connected" to another component does it mean that there is no other component between the component and the other component. The terms "first," "second," "third," etc., are only used to describe different components and do not limit the components themselves. Therefore, the first component can also be referred to as the second component. Moreover, the combination of components / units / circuits in this document is not a combination that is generally known, conventional, or customary in this field. Whether the component / unit / circuit itself is customary cannot be used to determine whether its combination relationship is easily completed by someone with ordinary knowledge in the art.

[0018] Please refer to Figures 1 and 2. Figure 1 is a block diagram illustrating the insect-borne sound wave detection system 100 according to the first embodiment of the present invention; Figure 2 is a schematic diagram illustrating the use of the insect-borne sound wave detection system 100 according to Figure 1. The insect-borne sound wave detection system 100 includes a sound receiving device 110, a storage device 120, and a processing device 130. The sound receiving device 110 is used to receive sound from an object 10 to generate a sound signal (see Figure 3). The storage device 120 is signal-connected to the sound receiving device 110 and stores the sound signal. The processing device 130 is signal-connected to the storage device 120 and receives the sound signal. The processing device 130 includes a feature extraction module 131 and an artificial intelligence module 132. The feature extraction module 131 extracts multiple feature information from the sound signal. Artificial intelligence module 132 connects to feature extraction module 131 via signal, inputs these feature information into a first identification model M1 to generate a first identification result, and inputs these feature information into a second identification model M2 to generate a second identification result. The first identification result includes whether there is a borer in object 10, and the second identification result includes a category of borer.

[0019] In detail, the sound receiving device 110 may include a listening probe, a sensor encapsulating a surface acoustic wave filter, or a piezoelectric thin-film electronic stethoscope; the storage device 120 may be a memory or other storage device; the processing device 130 may be a microprocessor, a central processing unit (CPU), or an edge computing device; the object 10 may be a tree, a cultural relic made of wood, a sculpture, or other wooden item, but the present invention is not limited thereto. As shown in Figure 2, the sound receiving device 110 is connected to the object 10 to record the sound in the object 10, and the sound signal can be recorded by an external recording module to generate an audio file, which is then stored in the storage device 120.

[0020] Further, the feature extraction module 131 acquires the sound signal from the storage device 120, preprocesses the sound signal, and extracts feature information from the sound signal. Either the first identification model M1 or the second identification model M2 may include a moving window converter. The first identification result includes whether there are wood-boring insects in the object 10 or not. The second identification result includes whether the wood-boring insects are beetles, longhorn beetles, termites, or other insect types, but the present invention is not limited thereto.

[0021] In detail, the moving window converter contains a hierarchical structure, which can process images of different scales and performs well in object detection and object segmentation. The first recognition model M1 and the second recognition model M2 may include a moving window self-attention mechanism (SWMSA), which restricts the computation of the self-attention mechanism to a local region, reduces computational complexity, and processes higher resolution feature information, thereby improving recognition accuracy.

[0022] Accordingly, the insect-boring sound detection system 100 of the present invention can effectively detect whether wooden articles contain the gnawing sounds of wood-boring insects through non-invasive detection, and identify the type of wood-boring insect by the characteristics of the sound, thereby avoiding damage to the wood or trees. The detailed operation of the feature extraction module 131 will be described below through a more detailed embodiment.

[0023] Please refer to Figures 1 and 3, wherein Figure 3 is a block diagram illustrating the insect-borne sound wave detection system 100a according to the second embodiment of the present invention. The insect-borne sound wave detection system 100a includes a sound receiving device 110, a storage device 120, and a processing device 130. The processing device 130 includes a feature extraction module 131a and an artificial intelligence module 132. In the second embodiment, the operation of the sound receiving device 110, the storage device 120, and the artificial intelligence module 132 of the processing device 130 is the same as that of the sound receiving device 110, the storage device 120, and the processing device 130 of the first embodiment, and will not be described again. In particular, the feature extraction module 131a may include a segmentation module 1311, a processing module 1312, a Fourier transform module 1313, a signal enhancement module 1314, and a feature conversion module 1315.

[0024] Please refer to Figures 3 and 4A together. Figure 4A is a schematic diagram illustrating how the segmentation module 1311 of the insect-boring sound wave detection system 100a according to Figure 3 segments the sound signal S1 into sound segment signals S2. The segmentation module 1311 divides the sound signal S1 into multiple sound segment signals S2. For example, the sound signal S1 recorded by the receiver 110 can be a sound waveform with a duration of 1 second, as shown in the upper half of Figure 4A. The segmentation module 1311 can divide the sound signal S1 into multiple sound segment signals S2 with a duration of 25 milliseconds.

[0025] Please refer to Figures 3, 4A, and 4B. Figure 4B is a schematic diagram of the segmented windowed signal S3 of the processing module 1312 of the insect-boring sound wave detection system 100a according to Figure 3. The processing module 1312 is connected to the segmentation module 1311 and performs window function processing on these sound segment signals S2 to generate complex segmented windowed signals S3. In detail, the processing module 1312 reduces the signals on both sides of each sound segment signal S2 in Figure 4A, retaining the signal in the middle, to generate the segmented windowed signal S3 shown in Figure 4B. This reduces the impact of abrupt changes at both ends of the sound segment signal S2 on the subsequent identification results. The Fourier transform module 1313 is connected to the processing module 1312 and performs a Fourier transform on these segmented windowed signals S3 to convert them from the time domain to the frequency domain.

[0026] Please refer to Figures 3, 5A and 5B together. Figure 5A is a schematic diagram showing the segment windowing signal S3 of the signal enhancement module 1314 of the insect-infested acoustic detection system 100a in Figure 3 before enhancement. Figure 5B is a schematic diagram showing the segment windowing enhancement signal S4 of the signal enhancement module 1314 of the insect-infested acoustic detection system 100a in Figure 3.

[0027] The signal enhancement module 1314 is connected to the Fourier transform module 1313 and enhances the complex amplitude of a high-frequency segment in the windowed segment signal S3 to generate a complex windowed segment enhancement signal S4. Figure 5A illustrates the windowed segment signal S3 after Fourier transform. Generally, the amplitude of a sound waveform at high frequencies is smaller than that at low frequencies. The signal enhancement module 1314 increases the amplitude intensity of the high-frequency segment of the windowed segment signal S3 to generate the windowed segment enhancement signal S4, thereby improving the clarity and intelligibility of the signal in the high-frequency segment and preventing the omission of subtle feature information related to insects due to the weak signal amplitude in the high-frequency segment. Specifically, the frequency range of the high-frequency segment enhanced by the signal enhancement module 1314 is determined based on whether the signal is distorted or generates high-frequency noise after enhancement. In other embodiments of the present invention, the segment windowing signal S3 can also be enhanced by other signal enhancement methods to facilitate subsequent feature information extraction, but the present invention is not limited thereto.

[0028] The feature conversion module 1315 connects to the signal enhancement module 1314 and converts the segmented enhanced signal S4 through a Mel-scale filter to generate these feature information. Specifically, the feature information can be filter bank feature information. Compared with other feature extraction techniques, the Mel-scale filter can retain more information and has better adaptability to artificial intelligence models. Therefore, the insect-boring sound wave detection system 100a of the present invention can preprocess the sound signal S1 through the feature extraction module 131 to improve the accuracy and efficiency of feature extraction and detection.

[0029] Please refer to Figures 1, 2, and 6, wherein Figure 6 is a block diagram illustrating the insect-borne sound wave detection system 100b of the third embodiment of the present invention. The insect-borne sound wave detection system 100b includes a sound receiving device 110, a storage device 120, and a processing device 130. The processing device 130 includes a feature extraction module 131 and an artificial intelligence module 132. In the second embodiment, the feature extraction module 131 and artificial intelligence module 132 of the sound receiving device 110, storage device 120, and processing device 130 can operate in the same way as the feature extraction module 131a and artificial intelligence module 132 of the sound receiving device 110, storage device 120, and processing device 130 of the first embodiment of the insect-borne sound wave detection system 100, or the feature extraction module 131a and artificial intelligence module 132 of the sound receiving device 110, storage device 120, and processing device 130 of the second embodiment of the insect-borne sound wave detection system 100a, and will not be described again. In particular, the processing device 130 may further include a collection module 133. The collection module 133 is signal-connected to the artificial intelligence module 132, and marks the first identification result and the second identification result corresponding to the object 10 on the sound signal (see Figure 3) and stores them in the storage device 120.

[0030] In other words, after the artificial intelligence module 132 generates the first identification result and the second identification result through the first identification model M1 and the second identification model M2, it can collect the sound signals corresponding to different types of woodworms in the storage device 120 for subsequent woodworm sound wave identification.

[0031] Please refer to Figures 1, 2, and 7. Figure 7 is a flowchart illustrating the insect-borne sound wave detection method 200 according to the fourth embodiment of the present invention. The insect-borne sound wave detection method 200 includes steps 210, 220, 230, 240, and 250. Step 210 includes driving a sound receiving device 110 to receive sound from an object 10 to generate a sound signal (see Figure 3). Step 220 includes driving a storage device 120 to store the sound signal. Step 230 includes driving a feature extraction module 131 of a processing device 130 to extract multiple feature information from the sound signal. Step 240 includes driving an artificial intelligence module 132 of the processing device 130 to input these feature information into a first recognition model M1 to generate a first recognition result. Step 250 includes driving the artificial intelligence module 132 of the processing device 130 to input these feature information into a second recognition model M2 to generate a second recognition result. The first identification result includes whether there is a wood-boring insect in the object 10, and the second identification result includes a type of wood-boring insect. Therefore, the wood-boring acoustic detection method 200 of the present invention can not only detect damage to wood or trees early, but also eliminate wood-boring insects by targeting the specific type of insect, which is of great significance for preventing economic losses and protecting cultural assets.

[0032] As can be seen from the above embodiments, the insect-boring sound detection system and method of the present invention have the following advantages: First, it can effectively detect whether wooden items contain the gnawing sound of woodworms through non-invasive detection, and identify the type of woodworm by the characteristics of the sound, thereby avoiding damage to the wood or trees; Second, it can process the sound signal through a feature extraction module to improve the accuracy and efficiency of feature extraction and detection; Third, it can not only detect damage to wood or trees early, but also eliminate woodworms by targeting the type of woodworm, which is of great significance for preventing economic losses and protecting cultural assets.

[0033] Although the present invention has been disclosed above by way of embodiments, it is not intended to limit the present invention. Any person skilled in the art can make various modifications and refinements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims. [Simplified Explanation of the Diagram]

[0034] Figure 1 is a block diagram illustrating the insect-infested acoustic detection system of the first embodiment of the present invention; Figure 2 is a schematic diagram illustrating the use of the insect-infested acoustic detection system according to Figure 1; Figure 3 is a block diagram illustrating the insect-infested acoustic detection system of the second embodiment of the present invention; Figure 4A is a schematic diagram illustrating the segmentation module of the insect-infested acoustic detection system according to Figure 3 dividing the sound signal into sound segment signals; Figure 4B is a schematic diagram illustrating the segment windowing signal of the processing module of the insect-infested acoustic detection system according to Figure 3; Figure 5A is a schematic diagram illustrating the segment windowing signal before enhancement by the signal enhancement module of the insect-infested acoustic detection system according to Figure 3; Figure 5B is a schematic diagram illustrating the segment windowing enhancement signal of the signal enhancement module of the insect-infested acoustic detection system according to Figure 3. Figure 6 is a block diagram illustrating the insect-boring acoustic detection system of the third embodiment of the present invention; and Figure 7 is a flowchart illustrating the insect-boring acoustic detection method of the fourth embodiment of the present invention.

Claims

1. A sound wave detection system for insect infestation, comprising: a sound receiving device for receiving sound from an object to generate a sound signal; a storage device signal-connected to the sound receiving device and storing the sound signal; and a processing device signal-connected to the storage device and receiving the sound signal, the processing device comprising: a feature extraction module for extracting multiple feature information from the sound signal; and an artificial intelligence module signal-connected to the feature extraction module for inputting the feature information into a first recognition model to generate a first recognition result, and inputting the feature information into a second recognition model to generate a second recognition result; wherein... The first identification result includes whether there is a borer in the object, and the second identification result includes a category of the borer; wherein, the feature extraction module includes: a segmentation module, which segments the sound signal into multiple sound segment signals; a processing module, which is connected to the segmentation module and performs a window function processing on the sound segment signals to generate multiple segment windowed signals; a Fourier transform module, which is connected to the processing module and performs a Fourier transform on the segment windowed signals to convert them from the time domain to the frequency domain; and a feature transformation module, which is connected to the Fourier transform module and transforms the segment windowed signals through a Mel-scale filter to generate the feature information.

2. The insect-boring sound wave detection system as claimed in claim 1, wherein either the first identification model or the second identification model includes a shifted window transformer (Swin Transformer).

3. The insect-boring sound wave detection system as described in claim 1, wherein the sound receiving device includes a listening probe, a sensor encapsulating a surface acoustic wave filter, or a piezoelectric thin-film electronic stethoscope.

4. The insect-boring sound wave detection system as described in claim 1, wherein the processing device further comprises: a collection module, signal-connected to the artificial intelligence module, marking the first identification result and the second identification result corresponding to the object on the sound signal and storing them in the storage device.

5. A method for detecting insect infestation using sound waves, comprising: driving a sound receiving device to receive sound from an object to generate a sound signal; driving a storage device to store the sound signal; driving a feature extraction module of a processing device to extract multiple feature information from the sound signal; driving an artificial intelligence module of the processing device to input the feature information into a first recognition model to generate a first recognition result; and driving the artificial intelligence module of the processing device to input the feature information into a second recognition model to generate a second recognition result; wherein, The first identification result includes whether there is a woodworm in the object, and the second identification result includes a category of the woodworm; wherein, the woodworm sound wave detection method further includes: a segmentation module driving the feature extraction module to segment the sound signal into a complex number of sound segment signals; a processing module driving the feature extraction module to perform a window function processing on the sound segment signals to generate complex segment windowed signals; a Fourier transform module driving the feature extraction module to perform a Fourier transform on the segment windowed signals to convert them from the time domain to the frequency domain; and a feature transformation module driving the feature extraction module to transform the segment windowed signals through a Mel-scale filter to generate the feature information.

6. The insect-boring sound wave detection method as described in claim 5, wherein either the first identification model or the second identification model includes a moving window converter.

7. The method for detecting insect infestation sound waves as described in claim 5, wherein the sound receiving device includes a listening probe, a sensor encapsulating a surface acoustic wave filter, or a piezoelectric thin-film electronic stethoscope.

8. The insect-boring sound wave detection method as described in claim 5 further includes: driving a collection module of the processing device to mark the first identification result and the second identification result corresponding to the object on the sound signal and store them in the storage device.

Images