Parachute parameter identification method based on monocular video

Abstract

The invention provides a parachute parameter identification method based on a monocular video. Firstly, parachute video key frames are extracted, and a parachute image set is obtained; secondly, acquiring a connected domain covering a canopy through an image processing method, extracting a minimum rectangular bounding box of the connected domain, and cutting a parachute image to realize coarse positioning of a canopy area; thirdly, preprocessing the cut image to generate a grey-scale map suitable for segmentation, and completing accurate segmentation of a canopy area in combination with a threshold segmentation algorithm and a morphological processing method; meanwhile, a canopy top hole is introduced to serve as a reference object, a parachute inflation process deformation theoretical model is constructed, and scale factors of different sections of a canopy are calculated; and finally, solving to obtain a top hole coordinate, a canopy deflection angle and a canopy projection area by using a camera imaging model. According to the method, the utilization value of parachute video data is improved, and the parameters obtained through identification have important theoretical significance and practical value for parachute theoretical research and engineering design.

Description

Technical Field

[0001] The invention relates to the field of parachutes for spacecraft, and in particular to a parachute parameter identification method based on monocular video. Background Art

[0002] As a pneumatic deceleration device, parachutes are widely used in spacecraft recovery and landing processes because of their simple structure, efficient deceleration, reliable operation, and low cost. The working process of the parachute system is extremely complex, and its performance directly affects the safety and reliability of space missions. There are many cases in history where missions failed due to parachute system failures. Therefore, it is very important to analyze the performance of parachutes. The analysis and evaluation methods of parachute performance can be divided into experimental methods and simulation analysis methods. Among them, the airdrop test method is widely used because it can restore the actual working conditions of the parachute to a greater extent.

[0003] The relevant parameters of parachutes in airdrop tests are usually obtained by mechanical measurement, electrical measurement, optical measurement and other means. The optical measurement method is a non-contact measurement method. The working process of the parachute is recorded by optical equipment, and information is extracted from the acquired optical images or video data using image measurement technology. Compared with the other two methods, there is no need to install sensors on the canopy or parachute ropes during the measurement process, which will not affect the working state and dynamic characteristics of the parachute. Therefore, the optical measurement method is increasingly valued in the development of parachutes.

[0004] At present, parachute parameter identification is mostly performed using videos shot by binocular or multi-camera cameras, and the cameras need to be calibrated. However, due to the limitations or constraints of the test conditions, the videos obtained in the airdrop test can generally only be used as monocular videos for parameter identification, and the image resolution is limited, and there is a lack of calibration information such as the camera's in-camera posture. Therefore, the parachute video data currently obtained is mostly used for recording, storage and qualitative analysis, and is rarely used for quantitative measurement of parachute parameters.

[0005] In order to realize the identification of more types of parachute parameters based on existing video data and improve the parameter identification accuracy, the present invention proposes a parachute parameter identification method based on monocular video, which is suitable for the video on the parachute cabin taken by a monocular camera that has not been calibrated. The method realizes the identification of the canopy deflection angle, the coordinates of the canopy top hole and the canopy projection area. The identified parachute parameters can provide support for the establishment of parachute theoretical models, parachute performance analysis and engineering design. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention proposes a parachute parameter identification method suitable for parachute cabin video taken by a monocular camera without calibration. The method steps of the present invention are as follows:

[0007] Step 1: Based on the video of the parachute cabin taken by an uncalibrated monocular camera, the key frames of the parachute video are extracted to obtain a parachute video image set.

[0008] Step 2: Rough positioning of the canopy area.

[0009] (1) Based on the parachute image obtained in step 1, a connected domain covering the canopy area is obtained by image processing method.

[0010] (2) The minimum rectangular bounding box of the connected domain is extracted and the canopy image is cropped using the rectangular boundary to achieve coarse positioning of the canopy area and improve the image signal-to-noise ratio.

[0011] Step 3: Accurately segment the canopy area.

[0012] (1) The parachute cropped image in step 2 is preprocessed to obtain a grayscale image with obvious grayscale values ​​between the canopy area and other areas, so as to efficiently separate interference factors such as the sky background area, the parachute rope area, and the bifurcation belt area.

[0013] (2) The canopy area is segmented using the threshold segmentation algorithm, and the segmentation accuracy of the canopy area is improved through digital morphological processing. The number of pixels contained in the canopy area after accurate segmentation is calculated to be m, which lays the foundation for the identification of the canopy projection area.

[0014] Step 4, solve the proportional factors corresponding to different canopy cross-sections.

[0015] (1) Solve the proportional factor k1 corresponding to the plane of the top hole of the canopy. Select the top hole of the canopy as a reference, extract the outer contour of the top hole of the canopy in the image, and count the number of pixels occupied by the outer contour as n. It is known that the actual circumference of the top hole of the canopy is l a , solve to get the proportional factor k1 corresponding to the canopy top hole plane:

[0016] k1=la / n

[0017] (2) The camera coordinate system and image coordinate system of the parachute system are established, and the conclusion is drawn based on the camera imaging model: for two different planes parallel to the imaging plane, the corresponding scale factor is proportional to the distance between the two planes and the optical center of the camera.

[0018] (3) Establish a theoretical model of the canopy deformation during the parachute deployment process, and solve the distances from the canopy top hole plane and the canopy maximum projection cross section to the camera optical center, respectively: a and h p , solve to get the proportional factor k2 corresponding to the maximum projection cross section of the canopy:

[0019] k2=(h p / h a )k1

[0020] Step 5: parachute parameter identification.

[0021] (1) Solving the coordinates of the canopy top hole:

[0022] a. Based on the accurate segmentation of the canopy area obtained in step 3, the centroid coordinates of the canopy area are used as the coordinates of the canopy top hole in the image coordinate system (u p ,v p ):

[0023]

[0024] Where (u i ,v i ),i=1, … ,m is the coordinate of the pixel point in the canopy area.

[0025] b. According to the proportional factor between the camera imaging model and the top hole plane, solve the coordinates of the canopy top hole in the camera coordinate system (X p ,Y p ,Z p ):

[0026]

[0027] Where (u0, v0) is the coordinate of the center point of the image in the image coordinate system.

[0028] (2) Solution for canopy deflection angle:

[0029] a. In the imaging plane, the distance l between the canopy top hole in the image and the camera optical axis op (in pixels) is:

[0030]

[0031] b. In the camera coordinates, the distance between the top hole of the canopy and the camera optical axis is l st for:

[0032] l st =k1l op

[0033] c. Solve the canopy deflection angle θ:

[0034]

[0035] (3) Solve for the canopy projection area:

[0036] a. Solve the projection area of ​​the canopy relative to the camera optical axis:

[0037]

[0038] b. Solve for the projection area of ​​the canopy relative to the central axis of the canopy:

[0039]

[0040] The present invention realizes the separation of the canopy area and the interference area such as sunlight through the method of image cropping, improves the signal-to-noise ratio of the image, and lays the foundation for the accurate segmentation of the canopy. By preprocessing the cropped image, the direct separation of interference factors such as the parachute rope, the rotary joint and the sling is realized, and the efficiency of canopy segmentation is improved. Since the canopy top hole is not blocked and the shape remains unchanged during the parachute deployment process, the canopy top hole is selected as a reference to solve the proportional factor, which improves the stability of the proportional factor solution. Since the canopy top hole is located at the special position of the top of the parachute, the calculation process of the proportional factor solution and the parachute parameter identification is more concise. The center of the accurately segmented canopy area is used as the top hole center, which improves the efficiency of parachute parameter identification. The method provides a parachute parameter identification method based on monocular video, which is suitable for the video on the parachute cabin taken by an uncalibrated monocular camera. The identification process is simple and efficient, the identification result is accurate, and the utilization rate of the existing parachute video data is effectively improved. The identified parameters provide data support for parachute performance analysis and engineering design. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a block diagram of the overall concept of the present invention.

[0042] Figure 2 This is an example image of a parachute.

[0043] Figure 3 Cropped image of a parachute.

[0044] Figure 4 Accurately segment images for parachutes.

[0045] Figure 5 It is the sub-pixel outer contour image of the top hole of the canopy.

[0046] Figure 6 Schematic diagram of the camera coordinate system and image coordinate system in the parachute system.

[0047] Figure 7 Schematic diagram of the theoretical model of canopy deformation during the parachute deployment process.

[0048] Figure 8 This is a schematic diagram of the center position of the top hole of the canopy.

[0049] Fig. 9 Schematic diagram of the geometric relationship of the canopy deflection angle. DETAILED DESCRIPTION

[0050] The present invention is described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that this example is only used to illustrate the present invention and is not used to limit the scope of the present invention. The overall idea of ​​the parachute parameter identification method based on monocular video provided by the present invention is as follows: Figure 1 As shown, the specific implementation method includes the following steps:

[0051] Step 1: Based on the video of the parachute cabin shot by the uncalibrated monocular camera, the key frames of the parachute video are extracted to obtain the parachute video image set. Taking the 28606th frame of a parachute airdrop video as an example, the subsequent processing of the parachute image is explained. The example parachute image is as follows: Figure 2 shown.

[0052] Step 2: Rough positioning of the canopy area.

[0053] (1) Based on the parachute image obtained in step 1, a connected domain covering the canopy area is obtained by an image processing method. In order to eliminate the interference of sunlight in the background of the parachute image in this example, the connected domain is extracted in the HSV color space.

[0054] (2) Extract the minimum rectangular bounding box of the connected domain and use the rectangular boundary to crop the canopy image, achieve coarse positioning of the canopy area, and improve the image signal-to-noise ratio. The cropped image of the parachute is shown in the figure below: Figure 3 shown.

[0055] Step 3: Accurately segment the canopy area.

[0056] (1) The parachute cropped image in step 2 is preprocessed to obtain a grayscale image with obvious grayscale values ​​between the canopy area and other areas, so as to efficiently separate interference factors such as the sky background area, the parachute rope area, and the bifurcation belt area. In this example, in order to suppress the influence of uneven illumination and a small amount of noise in the cropped image, the grayscale image is generated based on the characteristics of the Lab color space and the RGB color space.

[0057] (2) The Otsu algorithm in the threshold segmentation algorithm is used to complete the segmentation of the canopy area, and the segmentation accuracy of the canopy area is improved through digital morphological processing. The number of pixels contained in the canopy area after accurate segmentation is counted as m, which lays the foundation for the identification of the canopy projection area. The accurate segmentation result of the parachute canopy area is shown in the figure. Figure 4 shown.

[0058] Step 4, solve the proportional factors corresponding to different canopy cross-sections.

[0059] (1) Solve the scale factor k1 corresponding to the plane of the top hole of the canopy. The top hole of the canopy is selected as a reference, and the outer contour of the top hole of the canopy in the image is extracted. In order to improve the accuracy of the scale factor calculation, this example uses a sub-pixel edge extraction algorithm to complete the extraction of the top hole outer contour. The sub-pixel outer contour of the top hole of the canopy is as follows: Figure 5 shown.

[0060] (2) The number of pixels occupied by the outer contour is counted as n, and the actual circumference of the top hole of the canopy is known to be l a , solve to get the proportional factor k1 corresponding to the canopy top hole plane:

[0061] k1=la / n

[0062] (3) Establish the camera coordinate system and image coordinate system of the parachute system, such as Figure 6 As shown, according to the camera imaging model, it is concluded that for two different planes parallel to the imaging plane, the corresponding scale factors are proportional to their distances from the optical center of the camera.

[0063] (4) Establish a theoretical model of the canopy deformation during the parachute deployment process, such as Figure 7 As shown in the figure, the distances from the canopy top hole plane and the canopy maximum projection section to the camera optical center are h a and h p , and obtain the proportional factor k2 corresponding to the maximum projection cross-section of the canopy:

[0064] k2=(h p / h a )k1

[0065] Step 5: parachute parameter identification.

[0066] (1) Solving the coordinates of the canopy top hole:

[0067] a. Based on the accurate segmentation of the canopy area obtained in step 3, the centroid coordinates of the canopy area are used as the coordinates of the canopy top hole in the image coordinate system (u p ,v p ):

[0068]

[0069] Where (u i ,v i ),i=1, … , m is the coordinate of the pixel point in the canopy area, and the top hole center position is as follows Figure 8 shown.

[0070] b. According to the proportional factor between the camera imaging model and the top hole plane, solve the coordinates of the canopy top hole in the camera coordinate system (X p ,Y p,Z p ):

[0071]

[0072] Where (u0, v0) is the coordinate of the center point of the image in the image coordinate system.

[0073] (2) Solution for canopy deflection angle:

[0074] a. In the imaging plane, the distance l between the canopy top hole in the image and the camera optical axis op (in pixels) is:

[0075]

[0076] b. In the camera coordinates, the distance between the top hole of the canopy and the camera optical axis is l st for:

[0077] l st =k1l op

[0078] c. Solve the canopy deflection angle θ:

[0079]

[0080] The geometric relationship diagram of the canopy deflection angle is as follows Fig. 9 shown.

[0081] (3) Solve for the canopy projection area:

[0082] a. Solve the projection area of ​​the canopy relative to the camera optical axis:

[0083]

[0084] b. Solve for the projection area of ​​the canopy relative to the central axis of the canopy:

[0085]

[0086] Finally, parachute parameter identification based on monocular video was achieved.

Claims

1. A parachute parameter identification method based on monocular video, characterized in that: The following steps are involved: Step 1, based on the video of the parachute cabin taken by an uncalibrated monocular camera, extract the key frames of the parachute video to obtain a parachute video image set; Step 2, based on the parachute image obtained in step 1, a rough positioning of the canopy area is achieved by an image processing method; Step 3, based on the parachute cropped image obtained in step 2, after preprocessing the image, the canopy area is accurately segmented by an image segmentation algorithm to obtain the number of pixels and pixel coordinate information in the canopy area; Step 4, taking the canopy top hole as a reference, and establishing a canopy deformation theoretical model during the parachute deployment process, and solving the proportional factors corresponding to different canopy cross-sections; Step 5, based on the information obtained in steps 3 and 4, the coordinates of the canopy top hole, the canopy deflection angle, and the canopy projection area are identified.

2. The method according to claim 1, characterized in that Coarse positioning of the canopy area, including; a. Obtain the connected domain covering the canopy area through image processing algorithm; b. Extract the minimum rectangular bounding box of the connected domain; c. Use the rectangular boundary to crop the canopy image, achieve coarse positioning of the canopy area, and improve the image signal-to-noise ratio.

3. The method according to claim 1, characterized in that Accurate segmentation of the canopy area, including: a. Preprocess the cropped image to eliminate image noise and improve image quality; b. Convert the image into a grayscale image and adjust the grayscale values ​​of the canopy area and other areas to make them significantly different, so as to facilitate the separation of interference factors; c. Use threshold segmentation algorithm and morphological processing method to achieve accurate segmentation of the canopy area.

4. The method according to claim 1, characterized in that: The proportional factors corresponding to different canopy sections are solved, including: a. Introduce the top hole of the canopy as a reference, extract the outer contour of the top hole of the canopy, and extract the number of pixels occupied by the contour. Knowing the perimeter of the top hole of the canopy, solve and obtain the proportional factor corresponding to the plane of the top hole of the canopy, where the proportional factor corresponding to the plane of the top hole of the canopy is the ratio of the perimeter of the top hole of the canopy to the number of pixels occupied by the outer contour of the canopy; b. Establish the camera coordinate system and image coordinate system of the parachute system, and draw the conclusion based on the camera imaging model: for different planes parallel to the imaging plane, the corresponding scale factor is proportional to the distance between the two planes and the optical center of the camera; c. Establish a theoretical model of canopy deformation during the parachute deployment process, obtain the distance between the canopy top hole plane and the canopy maximum projection cross-section to the camera optical center, and calculate the proportional factor corresponding to the canopy maximum projection cross-section.

5. The method according to claim 1, characterized in that Parachute parameter identification, including: a. Taking the centroid of the accurately segmented canopy area as the position of the canopy top hole in the image, and solving the coordinates of the canopy top hole in the image coordinate system; b. Using the scale factor corresponding to the camera imaging model and the top hole plane, solve the coordinates of the canopy top hole in the camera coordinate system; c. Using the coordinates of the canopy top hole in the camera coordinate system, calculate the canopy deflection angle, where the canopy deflection angle is the angle between the canopy center axis and the camera optical axis, and the canopy center axis is defined as the line between the canopy top hole and the camera optical center; d. Using the proportional factor of the number of pixels in the precise segmentation area of ​​the canopy and the maximum projection cross-section of the canopy, the projection area of ​​the canopy relative to the camera optical axis is solved. On this basis, the canopy deflection angle is introduced to complete the projection area correction, and the projection area of ​​the canopy relative to the central axis of the canopy is solved.

Images