DMD-based local imaging method and spectrum system

Abstract

The invention discloses a DMD-based local imaging method and a spectrum system, and the method comprises the steps: S1, pre-constructing an instance segmentation model which is used for recognizing a target needing local imaging before imaging, and is trained by learning the individual features and boundary information of the target needing local imaging; s2, an imaging device carrying the instance segmentation model is adopted to carry out local shooting imaging on a target needing local imaging, in the shooting process, the imaging device drives a turnover assembly to turn over to a designated position according to the recognition result of the instance segmentation model, after the direction and intensity of output light beams are adjusted, the output light beams are focused after being split, and the target is obtained; the light of each wavelength can be clearly imaged, a local image of a target at any position can be accurately obtained, and the problems of low data acquisition efficiency and data redundancy caused by the fact that an existing spectrograph cannot carry out flexible scanning imaging on a specific area are solved.

Description

Technical Field

[0001] The present invention relates to the field of spectral technology, and in particular to a local imaging method and a spectral system based on DMD. Background Art

[0002] Spectrometers can analyze the composition and structure of materials by accurately measuring the absorption, reflection or emission characteristics of materials to light of different wavelengths. With its precise analysis capabilities, spectrometers are widely used in agricultural monitoring, environmental testing, medical diagnosis and other fields. At present, most spectrometers use push-scan imaging technology to scan the target area point by point in the vertical direction through multiple detector arrays to obtain hyperspectral data.

[0003] Existing push-broom spectrometers usually need to scan the entire target area to obtain hyperspectral data, but in actual applications, many scenarios only need to obtain data from specific areas. Taking the orchard scene as an example, researchers and fruit farmers usually only focus on the hyperspectral data of fruit tree leaves or fruits to monitor the growth status of fruit trees, pests and diseases, and evaluate the quality of fruits. However, when a spectrometer using push-broom imaging technology obtains specific target data, it will inevitably collect data from all areas along the entire scanning path, including a large amount of irrelevant soil, branches, air, etc.

[0004] This acquisition method greatly reduces efficiency. On the one hand, scanning the entire area is time-consuming, labor-intensive and energy-consuming, which greatly limits work efficiency when resources are limited; on the other hand, a large amount of irrelevant data leads to data redundancy. These redundant data not only occupy a large amount of storage resources and increase storage costs, but also significantly increase the amount of calculation in subsequent data processing and analysis, reducing analysis speed and accuracy. Therefore, it is urgent to develop an imaging system and method that can efficiently and accurately obtain hyperspectral data of a specific area. Summary of the invention

[0005] In view of the problems in the related art, the present invention proposes a local imaging method and a spectral system based on DMD to overcome the above technical problems existing in the existing related art.

[0006] To this end, the specific technical solution adopted by the present invention is as follows: A local imaging method based on DMD, comprising: S1: pre-build an instance segmentation model, the instance segmentation model is used to identify the target required for local imaging before imaging, and the instance segmentation model is trained by learning the individual features and boundary information of the required local imaging target; S2: An imaging device equipped with an instance segmentation model is used to perform local imaging of the target for which local imaging is required. During the shooting process, the imaging device drives the flip component to flip to the specified position according to the recognition result of the instance segmentation model, adjusts the direction and intensity of the output light beam, and then focuses the output light beam after splitting it, so that each wavelength of light can be clearly imaged and the local image of the target at any position can be accurately obtained.

[0007] Furthermore, the training of the instance segmentation model includes: S1.1: Select images of targets in different regions, different lighting conditions, different growth stages, and different weather conditions, ensuring that the images cover different forms of targets, including single targets, clustered targets, and targets of different sizes and colors; S1.2: annotate the acquired image with objects, wherein, for an image with more than one object, annotate each object individually; S1.3: Assign a category label to each annotated object; S1.4: Divide the labeled target images and their corresponding category labels into training set, validation set and test set according to a certain ratio to obtain the instance segmentation model.

[0008] Furthermore, the YOLOv11 deep neural network is used to train the instance segmentation model.

[0009] A DMD-based local imaging spectral system, used for the above-mentioned DMD-based local imaging method, comprising: Model building module: pre-building an instance segmentation model, the instance segmentation model is used to identify the target of the required local imaging before imaging, and the instance segmentation model is trained by learning the individual features and boundary information of the required local imaging target; Local imaging module: An imaging device equipped with an instance segmentation model is used to perform local imaging of the target required for local imaging. During the shooting process, the imaging device drives the flip component to flip to the specified position according to the recognition result of the instance segmentation model, adjusts the direction and intensity of the output light beam, and then focuses the output light beam after splitting, so that each wavelength of light can be clearly imaged and the local image of the target at any position can be accurately obtained. Furthermore, the spectral range of the spectrometer covers 200-800nm, and the FWHH resolution is less than 2nm.

[0010] Furthermore, the imaging device includes an imaging lens, a beam splitter, a visible light camera, a rotatable reflector, a first collimating lens group, a cosine corrector, a DMD, a second collimating lens group, a grating, a focusing lens group and an imaging element. The imaging lens is installed at the starting position of the system optical path, the beam splitter is installed in the focusing optical path of the imaging lens, the beam splitter is in the horizontal direction, and its center is flush with the optical axis of the imaging lens and DMD, the beam splitter is in the vertical direction, and its center is flush with the optical axis of the visible light camera, the rotatable reflector is installed between the beam splitter and the DMD, the visible light camera is set above the beam splitter, the visible light camera is installed on the optical path in the vertical direction of the beam splitter, the optical center coincides with the output optical axis in the vertical direction of the beam splitter, the optical center of the rotatable reflector in the vertical direction is flush with the optical axes of the first collimating lens group and the cosine corrector, the cosine corrector is located above the first collimating lens group, the top of the cosine corrector is exposed outside the imaging device, the DMD is installed on one side of the rotatable reflector at a certain angle horizontally, and the second collimating lens The group, grating, focusing lens group and imaging element are arranged above the DMD in sequence, the optical axis coincides with the output optical path of the DMD, the visible light camera, the rotatable reflector and the imaging element are electrically connected to the main control board, the visible light camera receives the instructions of the main control board and uploads the image data, the main control board sends a command to control whether the rotatable reflector is started, when it is started, it can just allow the sunlight coming in from the cosine corrector to be reflected to the DMD after being collimated, when it is not started, it completely blocks the sunlight and allows the light output in the horizontal direction of the beam splitter to reach the DMD, the main control board outputs an instruction to control the flipping of the DMD, and ensures that when the galvanometer is not flipped on the optical path, the light signal output by the beam splitter cannot enter the subsequent collimation beam splitting optical path; when the galvanometer is flipped, the light signal output in the horizontal direction by the beam splitter can just be reflected to reach the subsequent collimation beam splitting optical path.

[0011] Furthermore, the wavelength response range of the imaging lens is adjusted according to the application scenario.

[0012] Furthermore, an electronic shutter is provided on one side of the optical path transmission of the imaging lens, and the switch of the electronic shutter controls whether the optical signal passes through the imaging lens.

[0013] Furthermore, the beam splitter splits the light signal image captured by the imaging lens into two beams before focusing. The light intensity ratio of the two beams can be 1:1, 2:8, 4:6 or any other ratio that meets the usage requirements. The center of the beam splitter is aligned with the optical axis of the imaging lens and the DMD in the horizontal direction, and is aligned with the optical axis of the visible light camera in the vertical direction.

[0014] Furthermore, the main control board includes at least one computing chip for coordinating the operation of the system, and the main control board transmits data with the computer via a cable or Bluetooth.

[0015] Furthermore, the grating, focusing lens group and imaging element are implemented by a single-point spectrometer, and the light inlet of the single-point spectrometer is installed at the focal position of the second collimating lens group to ensure that the maximum light flux of the incident light is transmitted to the optical fiber and then split. The spectrometer contains a collimating lens, a grating splitting system, a focusing lens and a linear array CCD imaging system, which can convert the collected spectral signal into digital data and complete the precise measurement of the spectral information of a specific point.

[0016] The beneficial effects of the present invention are: 1. This application solves the problem of low data collection efficiency and data redundancy caused by the inability of existing spectrometers to flexibly scan and image specific areas; 2. This application introduces an automatic calibration function, which enables the system to complete the calibration process more concisely without manual intervention. Automatic calibration can effectively ensure the accuracy and consistency of data collection, reduce the complexity of operation and the impact of human errors, and improve the stability and reliability of the system; 3. The present application adopts the DMD array column-by-column flipping method to replace the traditional grating scanning method, which effectively avoids the influence of the vibration that may be generated when the spectrometer performs line scanning on the fine optical path. The column-by-column flipping method of the DMD array can accurately control the projection of light, ensure the stability of the optical path during the scanning process, and further improve the accuracy of spectral data acquisition and the overall performance of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 is a structural schematic diagram of a spectral system based on local imaging of DMD according to Embodiment 2 of the present invention; Figure 2 is a schematic diagram of the position of a rotatable reflector when the DMD-based local imaging spectroscopy system according to the second embodiment of the present invention is not started; Figure 3 Schematic diagram of the position of the rotatable reflector when the DMD-based local imaging spectroscopy system according to the second embodiment of the present invention is started.

[0019] In the figure: 1. Imaging lens; 2. Electronic shutter; 3. Spectrum splitter; 4. Visible light camera; 5. Rotatable reflector; 6. First collimating lens group; 7. Cosine corrector; 8. DMD; 9. Second collimating lens group; 10. Grating; 11. Focusing lens group; 12. Imaging element; 13. Main control board; 14. Computer. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Embodiment 1

[0021] A local imaging method based on DMD, comprising: S1: Pre-build an instance segmentation model. The instance segmentation model is used to identify the target of the required local imaging before imaging. The instance segmentation model is trained by learning the individual features and boundary information of the required local imaging target. Specifically, taking the detection of bayberry in an orchard as an example, the training of the instance segmentation model includes: S1.1: Take enough images of bayberry in different areas of the orchard, under different lighting conditions, at different growth stages, and under different weather conditions to ensure that the images cover different forms of the target, including single bayberry, clustered bayberry, and bayberry of different sizes and colors; S1.2: For each bayberry image, carefully outline the outline of each bayberry to ensure the accuracy and completeness of the annotation. Each bayberry instance should be individually annotated, even if the bayberries are clustered, they should be annotated separately so that the model can learn the individual characteristics and boundary information of the bayberries. S1.3: Assign a category label to each annotated bayberry; S1.4: Divide the annotated bayberry images and their corresponding category labels into training set, validation set and test set according to a certain ratio. Use YOLOv11 deep neural network to train the instance segmentation model. The common division ratio is training set: validation set: test set = 7:2:1, but it can also be adjusted according to actual conditions. S2: An imaging device equipped with an instance segmentation model is used to perform local imaging of the target for which local imaging is required. During the shooting process, the imaging device drives the flip component to flip to the specified position according to the recognition result of the instance segmentation model, adjusts the direction and intensity of the output light beam, and then focuses the output light beam after splitting it, so that each wavelength of light can be clearly imaged and the local image of the target at any position can be accurately obtained. Embodiment 2

[0022] Taking the bayberry garden as a specific application scenario, detecting the sugar content or disease of bayberries still on the tree based on hyperspectral data, a spectral system based on local imaging of DMD is implemented by using the local imaging method based on DMD of embodiment 1, including: Model building module: pre-building an instance segmentation model, the instance segmentation model is used to identify the target of the required local imaging before imaging, and the instance segmentation model is trained by learning the individual features and boundary information of the required local imaging target; Local imaging module: An imaging device equipped with an instance segmentation model is used to perform local imaging of the target required for local imaging. During the shooting process, the imaging device drives the flip component to flip to the specified position according to the recognition result of the instance segmentation model, adjusts the direction and intensity of the output light beam, and then focuses the output light beam after splitting it, so that each wavelength of light can be clearly imaged and the local image of the target at any position can be accurately obtained.

[0023] like Figure 1-3 As shown, the imaging device includes an imaging lens 1, a beam splitter 3, a visible light camera 4, a rotatable reflector 5, a first collimating lens group 6, a cosine corrector 7, a DMD 8, a second collimating lens group 9, a grating 10, a focusing lens group 11 and an imaging element 12. The imaging lens 1 is installed at the starting position of the system optical path. The wavelength response range of the imaging lens 1 is adjusted according to the application scenario. The beam splitter 3 is installed in the focusing optical path of the imaging lens 1. The beam splitter 3 is in the horizontal direction. The beam splitter 3 splits the optical signal image captured by the imaging lens 1 into two beams before focusing. The light intensity ratio of the two beams is adjusted according to the use requirements. For example, the light intensity ratio of the two beams can be 1:1, 2:8, 4:6 or any other ratio that meets the use requirements. Its center in the horizontal direction is flush with the optical axis of the imaging lens 1 and the DMD 8, and in the vertical direction, its center is flush with the optical axis of the visible light camera 4. An electronic shutter 2 is provided on the optical path transmission side of the imaging lens 1. The switch of the electronic shutter 2 controls whether the optical signal passes through the imaging lens 1. The visible light camera 4 is arranged above the beam splitter 3. The visible light camera 4 is installed on the optical path in the vertical direction of the beam splitter 3. The optical center coincides with the output optical axis in the vertical direction of the beam splitter 3, and is connected to the main control board 12 through a cable for receiving instructions and uploading image data. The rotatable reflector 5 is installed between the beam splitter 3 and the DMD8. The rotatable reflector 5 is installed between the beam splitter 3 and the DMD8 in the horizontal direction. The optical center of the rotatable reflector 5 is flush with the optical axis of the first collimating lens group 6 and the cosine corrector 7 in the vertical direction, and is connected to the main control board 13 through a cable. The main control board 13 sends a command to control whether the rotatable reflector 5 is started. When started, it can just allow the sunlight coming in from the cosine corrector 7 to be reflected to the DMD8 after collimation. When not started, it completely blocks the sunlight and allows the light output in the horizontal direction of the beam splitter 3 to reach the DMD8. The cosine corrector 7 is located above the first collimating lens group 6, and the top of the cosine corrector 7 is exposed outside the imaging device. The DMD8 is installed on one side of the rotatable reflector 5 at a certain angle horizontally. The DMD8 is connected to the main control board 13 through a cable to receive the flip command. When the galvanometer is not flipped on the optical path, the optical signal output by the spectrometer 3 cannot enter the subsequent collimation spectrometer optical path; when the galvanometer is flipped, the optical signal output by the spectrometer 3 in the horizontal direction can just reach the subsequent collimation spectrometer optical path after reflection. The second collimating lens group 9, grating 10, focusing lens group 11 and imaging element 12 are arranged above DMD8 in sequence, and the optical axis coincides with the output optical path of DMD8, ensuring that the optical signal output by DMD8 can accurately enter the subsequent spectrometer imaging system. At the same time, the imaging element 12 is connected to the main control board 13 through a cable to transmit data to the main control board 13. The main control board 13 includes at least one computing power chip for coordinating the operation of the system. The main control board 13 transmits data with the computer 14 through a cable or Bluetooth.

[0024] It should be noted that, except for the computer 14, all the above-mentioned optical and electronic components are installed on the same set of high-strength and high-precision structural parts. The structural parts are made of high-quality materials, taking into account factors such as heat dissipation and anti-interference, to ensure the relative position stability of each component.

[0025] The imaging principle of the imaging device is as follows: like Figure 2 As shown, in the initial state of the system, the electronic shutter 2 is not pressed, and the rotatable reflector 5 is not started. At this time, no light signal passes through the optical path, and the system enters the dark background signal acquisition mode; like Figure 3 As shown, when the rotatable reflector 5 is started, the system enters the ambient light collection mode, and the external light source such as sunlight is corrected for uniform light intensity by the cosine corrector 7 and collimated by the first collimating lens group 6. After being reflected by the reflector, the light beam will be transmitted to the DMD8. At this time, the main control board 13 accurately controls the DMD8 to ensure that the light source signal is correctly output to the subsequent optical path; After the electronic shutter 2 is pressed, the rotatable reflector 5 is in an inactive state, and the imaging lens 1 starts to capture the optical signal image of the measured area. The optical signal collected by the imaging lens 1 is split by the beam splitter 3 into two beams. One beam is collimated and focused on the visible light camera 4, and the other beam passes through the inactive rotatable reflector 5 and continues to be transmitted to the DMD 8 for subsequent optical processing.

[0026] The pixels of the image captured by the visible light camera correspond to one or several galvanometers of DMD8, and are combined with the pre-loaded instance segmentation model through the main control board 13 to perform regional recognition and analysis of the captured image in real time. According to the recognition result, the main control board 13 controls the DMD8 to perform precise pixel flipping operations, and adjusts the direction and intensity of the output light beam by flipping the pixels of the specified area on the DMD8 column by column. This process is further collimated by the second collimating lens group 9 to ensure that the light beam propagates in a predetermined direction. The light beam regulated by DMD8 enters the grating 10, and the light beam after the grating 10 splits it is precisely focused by the focusing lens group 11 to ensure that each wavelength of light can be clearly imaged on the imaging element 12; The imaging element 12 converts the focused optical signal into an electrical signal and transmits it to the main control board 13 for data processing and storage. The main control board 13 can be used to temporarily store data, or a model can be installed to further analyze and optimize the data according to a preset algorithm. The processed data can be transmitted to the computer 14 via a cable or Bluetooth for in-depth analysis or display. The computer 14 can issue a model algorithm, or receive data transmission and perform subsequent analysis. Embodiment 3

[0027] Compared with the second embodiment, the grating 10, the focusing lens group 11 and the imaging element 12 of this embodiment are implemented by a single-point spectrometer. The light inlet of the single-point spectrometer is installed at the focal position of the second collimating lens group 9 to ensure that the maximum light flux of the incident light is transmitted into the optical fiber and then split. The spectrometer contains a collimating lens, a grating splitting system, a focusing lens and a linear array CCD imaging system, which can convert the collected spectral signal into digital data and complete the accurate measurement of the spectral information of a specific point. The difference between the imaging method of this embodiment and the second embodiment is that: according to the recognition result, the main control board 13 controls the DMD8 to flip from left to right and from top to bottom in sequence, and the optical signal output by the DMD8 is focused on the optical fiber interface of the single-point spectrometer through the second collimating lens group 9. After the single-point spectrometer performs splitting, the data is transmitted to the main control board 13. The use of a single-point spectrometer can greatly reduce the equipment cost, and the single-point spectrometer uses a linear array detector, and the optimization difficulty is much lower than that of a planar array detector.

[0028] To summarize, the system instrument first confirms that the reflector is closed, collects a dark noise data when the shutter is not pressed, then opens the reflector, collects the spectral data of the sun's ambient light, and then collects data, presses the shutter, and the light is divided into two beams by the spectrometer. After visible light imaging, the visible light image transmits the photo to the main control board, which is equipped with an instance segmentation model to segment the target object from the image based on my pre-trained data. The target is bayberry, and the image of bayberry is segmented to obtain a mask or logic matrix, where only the target area is 1 and the rest are 0. The main control board controls the DMD flip, and the DMD galvanometer array corresponds to the visible light pixel. Only the target area is flipped, which is the part of 1. The subsequent flipping is column by column or several columns corresponding to one column, as a strategy to replace the slit. Finally, the spectral data is obtained by spectrometry and then imaging is performed.

[0029] The present invention is equipped with a flexible local scanning function, which can avoid the redundant data generated in the traditional full-field scanning, thereby significantly improving the scanning efficiency. This function enables the device to scan the local area accurately when needed, reduces unnecessary data collection, and optimizes the spectral data processing process, thereby saving time and computing resources.

[0030] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A local imaging method based on DMD, characterized in that: include: S1: pre-build an instance segmentation model, the instance segmentation model is used to identify the target required for local imaging before imaging, and the instance segmentation model is trained by learning the individual features and boundary information of the required local imaging target; S2: An imaging device equipped with an instance segmentation model is used to perform local imaging of the target for which local imaging is required. During the shooting process, the imaging device drives the flip component to flip to the specified position according to the recognition result of the instance segmentation model, adjusts the direction and intensity of the output light beam, and then focuses the output light beam after splitting it, so that each wavelength of light can be clearly imaged and the local image of the target at any position can be accurately obtained.

2. A DMD-based local imaging method according to claim 1, characterized in that: Training of the instance segmentation model involves: S1.1: Select images of targets in different regions, different lighting conditions, different growth stages, and different weather conditions, ensuring that the images cover different forms of targets, including single targets, clustered targets, and targets of different sizes and colors; S1.2: annotate the acquired image with objects, wherein, for an image with more than one object, annotate each object individually; S1.3: Assign a category label to each annotated object; S1.4: Divide the labeled target images and their corresponding category labels into training set, validation set and test set according to a certain ratio to obtain the instance segmentation model.

3. A DMD-based local imaging method according to claim 1, characterized in that: The instance segmentation model is trained using the YOLOv11 deep neural network.

4. A spectral system for local imaging based on DMD, used to implement a local imaging method based on DMD according to any one of claims 1 to 3, characterized in that: include: Model building module: pre-building an instance segmentation model, the instance segmentation model is used to identify the target of the required local imaging before imaging, and the instance segmentation model is trained by learning the individual features and boundary information of the required local imaging target; Local imaging module: An imaging device equipped with an instance segmentation model is used to perform local imaging of the target required for local imaging. During the shooting process, the imaging device drives the flip component to flip to the specified position according to the recognition result of the instance segmentation model, adjusts the direction and intensity of the output light beam, and then focuses the output light beam after splitting it, so that each wavelength of light can be clearly imaged and the local image of the target at any position can be accurately obtained.

5. A spectral system based on local imaging of DMD according to claim 4, characterized in that: The imaging device comprises an imaging lens (1), a beam splitter (3), a visible light camera (4), a rotatable reflector (5), a first collimating lens group (6), a cosine corrector (7), a DMD (8), a second collimating lens group (9), a grating (10), a focusing lens group (11), and an imaging element (12). The imaging lens (1) is installed at the starting position of the system optical path, the beam splitter (3) is installed in the focusing optical path of the imaging lens (1), the center of the beam splitter (3) is aligned with the optical axis of the imaging lens (1) and the DMD (8) in the horizontal direction, the center of the beam splitter (3) is aligned with the optical axis of the visible light camera (4) in the vertical direction, the rotatable reflector (5) is installed between the beam splitter (3) and the DMD (8), the visible light camera (4) is arranged above the beam splitter (3), and the optical center of the rotatable reflector (5) is aligned with the first collimating lens group (6) and the cosine correction lens group (8) in the vertical direction. The optical axis of the corrector (7) is aligned, the cosine corrector (7) is located above the first collimating lens group (6), the top end of the cosine corrector (7) is exposed outside the imaging device, the DMD (8) is installed horizontally at a certain angle on one side of the rotatable reflector (5), the second collimating lens group (9), the grating (10), the focusing lens group (11) and the imaging element (12) are arranged in sequence above the DMD (8), the optical axis coincides with the output optical path of the DMD (8), and the visible light camera (4), the rotatable reflector (5) and the imaging element (12) are electrically connected to the main control board (13).

6. A spectral system based on local imaging of DMD according to claim 5, characterized in that: The wavelength response range of the imaging lens (1) is adjusted according to the application scenario.

7. A spectral system based on local imaging of DMD according to claim 5, characterized in that: An electronic shutter (2) is provided on the optical transmission side of the imaging lens (1), and the switch of the electronic shutter (2) controls whether the optical signal passes through the imaging lens (1).

8. The spectral system based on local imaging of DMD according to claim 5, characterized in that: The beam splitter (3) splits the light signal image captured by the imaging lens (1) into two beams before focusing, and the light intensity ratio of the two beams is adjusted according to usage requirements.

9. The spectral system based on local imaging of DMD according to claim 5, characterized in that: The grating (10), the focusing lens group (11) and the imaging element (12) are implemented using a single-point spectrometer. The light inlet of the single-point spectrometer is installed at the focal position of the second collimating lens group (9). The spectrometer contains a collimating lens, a grating light splitting system, a focusing lens and a linear array CCD imaging system.

Images