CN112414375B

CN112414375B - Unmanned aerial vehicle image posture recovery method for flood disaster emergency quick jigsaw making

Info

Publication number: CN112414375B
Application number: CN202011068346.7A
Authority: CN
Inventors: 段延松; 张祖勋; 刘昆波; 赵新博
Original assignee: Wuhan University WHU
Current assignee: Wuhan University WHU
Priority date: 2020-10-08
Filing date: 2020-10-08
Publication date: 2021-09-03
Anticipated expiration: 2040-10-08
Also published as: CN112414375A

Abstract

The invention mainly aims at the shortage of the UAV photogrammetry technology in obtaining quick puzzles in flood disaster emergency, and proposes a UAV image attitude recovery method for flood disaster emergency quick puzzle production. The core idea of the present invention is to use GNSS positioning information to firstly partition all images according to the non-flooded ground surface and perform spatial triangulation respectively. Interpolate to get the complete pose parameters of all images. This method utilizes the characteristics of UAV's stable attitude and continuous shooting, and obtains high-precision attitude parameters of the image of the flooded area through interpolation, so as to finally obtain a complete quick puzzle of the disaster area, effectively making up for the use of UAV photogrammetry technology in floods and floods. Deficiencies in disaster response.

Description

Unmanned aerial vehicle image posture recovery method for flood disaster emergency quick jigsaw making

Technical Field

The invention belongs to the field of aerial photogrammetry facing flood disaster emergency, and relates to an unmanned aerial vehicle image posture recovery method facing flood disaster emergency fast jigsaw puzzle manufacturing, wherein full image posture recovery based on continuous time continuous interpolation is a key technology of the method.

Background

China is a country with a great and frequent geological disasters, and the direct loss caused by China each year reaches billions of yuan, thereby bringing great threat to the safety of people in disaster areas. After a disaster occurs, the overall situation of the disaster area is rapidly and accurately acquired, and the key for carrying out disaster emergency command and rescue is to grasp the disaster distribution. In recent years, due to the characteristics of flexibility, low cost, no time and space limitation, high positioning precision and stable posture of a carried Global Navigation Satellite System (GNSS) receiver, the unmanned aerial vehicle remote sensing platform (UAV) photogrammetry technology, particularly the unmanned aerial vehicle photogrammetry technology, is gradually applied and popularized in geological disaster investigation and emergency, and the quick jigsaw achievement of the unmanned aerial vehicle remote sensing platform also becomes one of the most important fine data sources for disaster emergency rescue.

However, in the emergency of flood disasters, the connection points of the image overlapping areas in the large-area water area range are almost difficult to obtain, so that large-area weak connection or even no connection area occurs when all images are subjected to aerial triangulation (air-triple) operation, and the unmanned aerial vehicle image in the whole disaster area is passively divided into a plurality of discontinuous sub-air-triple areas covering the land with a large area. The posture of the image in the flooding area cannot be recovered undoubtedly, the integrity of the quick jigsaw data in the disaster area is reduced, a disaster blind area is formed, and the uncertainty of emergency disaster relief is increased. Therefore, the full-image posture recovery and the complete and fast jigsaw generation of the unmanned aerial vehicle data are necessary requirements for flood emergency relief at present.

At present, the existing photogrammetry software and method can only process land containing a small amount of water areas (a single image is not a water area completely), and cannot be completely suitable for unmanned aerial vehicle data processing in flood disaster areas of large-range water areas. Therefore, when flood disaster data is processed, the existing processing flow firstly divides the affected area into sub-areas (generally, each sub-area is an independent area which is not flooded by water), then respectively carries out unmanned aerial vehicle data acquisition and three-attitude empty recovery processing, and finally forms a plurality of sub-quick puzzles. The processing flow can not obtain the fast splicing result of the flooded area, so that the integrity and the applicability of the fast splicing are reduced, and the uncertainty of emergency decision is increased.

Disclosure of Invention

The invention mainly aims at the defect that the unmanned aerial vehicle photogrammetry technology can quickly obtain jigsaw in flood disaster emergency, and provides an unmanned aerial vehicle image posture recovery method for quickly making jigsaw in flood disaster emergency. The method utilizes the characteristics of stable posture and continuous shooting of the unmanned aerial vehicle, obtains the posture parameters with higher precision of the images of the flooded area through interpolation according to the three results of the non-flooded area, and finally obtains the complete quick jigsaw of the affected area.

In order to obtain a complete and fast jigsaw puzzle in a flood disaster area, the invention provides a full-image posture recovery method based on time continuous interpolation. The method has the core idea that GNSS positioning information is utilized, all images are partitioned according to a non-water-flooded ground surface and are respectively subjected to space-time-space-time resolving, and then attitude interpolation is carried out on the images of the water-flooded area which do not participate in resolving by utilizing the characteristic of continuous image acquisition, so that complete attitude parameters of all the images are obtained. According to the method, the high-precision attitude parameters of the images of the flooded area are obtained through interpolation by utilizing the characteristics of stable attitude and continuous shooting of the unmanned aerial vehicle, so that the complete quick jigsaw puzzle of the disaster area is finally obtained, and the defects of the unmanned aerial vehicle photogrammetry technology in flood disaster emergency are effectively overcome.

The technical problem of the invention is mainly solved by the following technical scheme:

the method is used for manufacturing the unmanned aerial vehicle image fast jigsaw during flood disaster emergency, and comprises four main parts of disaster area complete data acquisition, subarea division, empty and empty image attitude interpolation, flooding image attitude interpolation and fast jigsaw generation, and the specific implementation steps are as follows:

step 1: and selecting a proper unmanned aerial vehicle with a high-precision GNSS receiver according to the disaster situation.

Step 2: the selected unmanned aerial vehicle is used for carrying out complete aerial photography on the disaster area, and the starting position and the ending position of the aerial photography range are required to have at least N₁The image covers the non-water-flooded area, and each image has more than 60% of the non-water-flooded area, generally N₁≥6；

And step 3: arranging all images according to the GNSS positions, and manually dividing three independent empty sub-areas including a non-water-flooded area by using software such as DPgrid and the like. For solution stability, each subregion should contain at least N₂Sheet image, generally N₂≥6。

And 4, step 4: and performing space-three calculation on each sub-area by using software such as DPgrid and the like to obtain attitude parameters of the images participating in calculation. It should be noted that, because all images are acquired by the same unmanned aerial vehicle and the information such as the flying height of each image at the exposure time is basically consistent, the focal length, the image principal point and the distortion parameter of the images in all sub-areas should be kept consistent during calculation. Specifically, a subregion with the largest number of images can be selected for parameter calculation, the calculated focal length, image principal point and distortion parameter of the camera are used as fixed input parameters of the remaining subregions, and then the spatial three calculation processing of the subregions is carried out.

And 5: and marking the image successfully solved by the attitude parameters as Known, and marking the image unsuccessfully solved and the image of the water-flooded area not participating in calculation as Unknown. And all images are sorted according to the generation time.

Step 6: as shown in equation (1), the image labeled as UnKnown is interpolated with the pose parameters. And the focal length, image principal point and distortion parameters of the image labeled Known are copied to the image labeled UnKnown.

Wherein Atti represents attitude parameters (heading angle, pitch angle, roll angle),

representing the pose parameters of the UnKnown image i to be solved,

and

representing the pose parameters of the first Known image Pre and Last, obtained by searching from image i using time forward and backward.

And

the data capturing times of the videos i, Pre, and Last are shown.

And 7: after the attitude parameters of all the images are calculated, the ground elevation is set, geometric correction is carried out on all the images, and a complete quick jigsaw is generated.

Compared with the prior art, the invention has the advantages and beneficial effects that:

in the prior art, the images of the flooded area cannot be automatically processed, the images of the flooded area must be repaired manually, time and labor are wasted, and the processing result is not guaranteed precisely. The invention fully utilizes the characteristic of continuous image acquisition and can fully automatically recover the attitude parameters of the image of the flooded area, thereby enabling the image of the flooded area to effectively participate in the automatic processing of the fast jigsaw, ensuring the fast full-automatic production of the emergency image map of the flood area and effectively making up the defects of the prior photogrammetry technology in the emergency of flood disasters.

Drawings

FIG. 1 is a general flow diagram of an embodiment of the present invention;

FIG. 2 is a graph showing gps location of Poyang lake water flooded area data according to the present invention;

FIG. 3 is a diagram illustrating the processing results of the present invention for a Poyang lake flooding area;

FIG. 4 is a diagram illustrating the processing result of the invention for a disaster-stricken area in Poyang lake.

Detailed Description

The technical solution of the present invention is described in detail below with reference to the accompanying drawings and examples.

The invention provides a technical scheme that an unmanned aerial vehicle image posture recovery method for flood disaster emergency quick jigsaw making is provided, wherein full image posture recovery based on continuous time continuous interpolation is a key technology of the invention.

Step 1: and selecting a proper surveying and mapping grade unmanned aerial vehicle with a high-precision GNSS receiver and a high-precision camera.

Step 2: the method comprises the steps that an unmanned aerial vehicle is selected to carry out complete aerial photographing flight on a disaster area, the starting position and the ending position of an aerial photographing range are required to at least comprise N images to cover non-water-flooded areas, and more than 60% of the non-water-flooded areas are arranged in the images; for solution stability, N is typically greater than 6.

And step 3: and arranging all the images according to the GNSS positions, and manually dividing independent empty three sub-areas containing non-water-flooded areas by using DPgrid or other empty three software. Each sub-area should contain at least 6 images.

And 4, step 4: and performing space-three calculation on each sub-area by using software such as DPgrid and the like to obtain attitude parameters of the images participating in calculation. During calculation, the focal length, the image principal point and the distortion parameter of all the images are required to be determined to be consistent. Specifically, the sub-regions with the largest number of images can be selected for parameter calculation, the calculated focal length, image principal point and distortion parameters of the camera are used as fixed input parameters of the remaining images, and then the space-three calculation processing of the sub-regions is carried out.

representing the attitude parameter of the UnKnown image i to be solved;

and

representing the attitude parameters of the first Known image Pre and Last obtained by searching forward and backward by using time from the image i;

and

the data capturing times of the videos i, Pre, and Last are shown.

And 7: and after the attitude parameters of all the images are solved, setting the ground elevation, and performing geometric correction on all the images to generate a complete quick jigsaw.

And 8: and outputting the result.

The technical effects of the present invention will be described with reference to the accompanying drawings, wherein fig. 2 is a diagram showing the gps position of the Poyang lake water flooded area data according to the present invention; after the above steps, fig. 3 is a processing result diagram of the Poyang lake flooding area, namely, the fast water area mosaic. Fig. 4 is a result diagram of application processing performed in 7-month Poyang lake extra-large flood disaster 2020, and the technology plays an important role in flood fighting and disaster relief in 7-month 2020, and is consistent and certain in national ministry of water conservancy, the water conservancy and profit living in Jiangxi province and the emergency disaster relief department in Jiangxi province.

The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.

Claims

1. an unmanned aerial vehicle image attitude recovery method for flood disaster emergency quick puzzle making is characterized in that, comprises the following steps:

Step 1: Select a suitable drone with a high-precision GNSS receiver according to the disaster situation;

Step 2: Use the selected drone to take complete aerial photography of the disaster area. It is required that there are at least N ₁ images covering the non-flooded area at the start and end positions of the aerial photography range, where N ₁ ≥ 6, and there are 60 images in each image. % or more of non-flooded areas;

Step 3: Arrange all the images according to the GNSS position, and divide into three independent empty sub-areas containing non-flooded areas; each sub-area should contain at least N ₂ images, of which N ₂ ≥ 6;

Step 4: Perform an air-triple calculation on each sub-area to obtain the attitude parameters of the images involved in the calculation;

When performing the spatial triangulation calculation for each sub-area, keep the focal length, image principal point and distortion parameters of the images in all sub-areas consistent, specifically: first select a sub-area with the largest number of images for parameter calculation, and calculate the calculated value. The camera focal length, image principal point and distortion parameters are used as the fixed input parameters of the remaining sub-regions, and then the space triangulation processing of these sub-regions is performed;

Step 5: Mark the images whose attitude parameters are successfully calculated as Known, and mark the images that have not been successfully calculated and the images of flooded areas that did not participate in the calculation as UnKnown, and sort all images according to the generation time;

Step 6: As shown in formula (1), perform attitude parameter interpolation on the image marked UnKnown, and copy the focal length, image principal point and distortion parameters of the image marked Known to the image marked UnKnown;

In the formula, Atti represents the attitude parameters, including the heading angle, pitch angle, roll angle,

Represents the attitude parameter of the UnKnown image i to be solved;

and

Represents the attitude parameters of the first Known image Pre and Last obtained by searching forward and backward in time from image i;

and

Indicates the data recording time of images i, Pre and Last;

Step 7: After the attitude parameters of all images are calculated, set the ground elevation, perform geometric correction on all images, and generate a complete quick puzzle.