Establishing the dependence of point position on number of stereo pairs based on data obtained from unmanned aerial vehicle

Тип работы:
Экономика и экономические науки


Детальная информация о работе

Выдержка из работы

Establishing the dependence of point position on number of stereo pairs based on data obtained from unmanned aerial vehicle
Section 2. Geodesy
Kuznetsova Oksana Pavlovna, master'-s student, Satpayev Kazakh National Technical University
Cathedra of Surveying and Geodesy E-mail: oksana_d_s@mail. ru Kuznetsova Irina Anatolevna, Ph. D., Associate Professor, Satpayev Kazakh National Technical University Cathedra of Surveying and Geodesy
E-mail: docent61@list. ru
Establishing the dependence of point position on number of stereo pairs based on data obtained from unmanned aerial vehicle
Abstract: This article considers the use of data obtained from a unmanned aerial vehicle for creation of orthophotoplans and digital topographic maps and plans. The dependence of point position-finding mean square errors on the number of stereo pairs is investigated.
Keywords: unmanned aerial vehicle, copter, tie points, stereo pair, correlation coefficient, aerial photography.
Today, unmanned aerial vehicles (UAVs or drones) are widely used worldwide — thanks to low cost, efficiency and versatility.
Dronescan effectively perform various tasks, including air surveillance and environmental monitoring, with subsequent processing and analysis of aerial photography.
Public services and commercial enterprises in many countries of the world now use unmanned systems for environmental monitoring [1, 115−118], design of geographic information systems, and other relevant purposes. In recent years, the use of drones is becoming increasingly popular in agriculture [2, 50−54] and in the forest sector [3, 97−99]. Aerial photo shoots obtained by unmanned systems are used for expeditious tracking of oil spills on land and water and for inspection of protected sections of oil and gas pipelines. Drones are also used in exploration of remote and extensive areas during emergency [4, 49−53].
There is a large number ofUAVs in the world. They can be divided into ultra-light, light, medium and heavy. They differ in specifications and set of characteristics (purpose, weight, size, flight duration and altitude, launching and landing system, autopilot and navigation systems, aerial photo and video format, etc.) [1, 115- 5, 27−31].
A drone is an aircraft without a human pilot aboard. It is used for reusable or conditionally reusable purposes.
It can move in the air independently and with a goal in mind to perform various functions in stand-alone mode (using its own control program) or remotely piloted by a human operator from a fixed or mobile control panel) [6, 17−18]. Copters are typical members of the unmanned aerial vehicles.
Copters are functionally similar to a helicopter. By number of rotors used (4, 6, 8), copters can be classified as quadcopter, hexocopter, etc. Modern controller mounted onboard the copter enables the aircraft to move in the air along a predetermined route that is set using specialized software [6, 61−68].
Airplane-type drones are used for capturing large areas, monitoring extended objects, and so on. There is often the need for shooting and monitoring small areas of land. To address this issue, it is expedient to use helicopter-type drones (copters), which have been applied in Kazakhstan.
An object located in the city ofAlmaty, Kazakhstan, was selected for shooting and further investigation. This object was 0. 3 km 2 in area. The territory was photographed using aerial quadcopter «AeroX». The external appearance is shown in picture 1. It was constructed by the National Center for Space Research and Technology. The photographs were taken from a height of 200 m- the size of one pixel is 16 cm.
Section 2. Geodesy
Picture 1. Aerial quadrcopter «AeroX»
Six routes covering the study area were marked out for the aerial photography. Obtained were 180 shots, with 90−92% fore-and-aft overlap and 30−35% lateral overlap.
The materials were processed using the software PHOTOMOD, in which four projects of one route to the same territory with different fore-and-aft overlaps were created. They are presented in picture 2.
Picture 2. Creating projects in PHOTOMOD program consisting of: a) 23 stereo pairs- b) 11 stereo pairs- c) 7 stereo pairs- d) 5 stereo pairs.
Before starting, it was established that the optimal number of tie points in one stereo pair in all the projects must be 23−25 tie points.
To create the first draft, as shown in picture 2a, consisting of a single route, all the photographs (24 photographs,
23 stereo pairs) were used. In this case, the fore-and-aft overlap was 90−92%. Identification obtained 623 points in 23 stereo pairs with a minimum correlation coefficient of 0,96. The total mean square error (MSE) in this case was 2,583, or 40 cm on ground (since one pixel is 16 cm in size).
Establishing the dependence of point position on number of stereo pairs based on data obtained from unmanned aerial vehicle
In the report in PHOTOMOD, errors on tie points between stereo pairs are equal to a pixel or a fraction of pixel. The value 0,2 m is automatically set as a predetermined permissible value of errors of determining the planned-high-altitude position of a point. This permissible error is set for processing aerospace photographs when the size of one pixel is 1m or more. In our case, the permissible error on tie points between stereo pairs can be set to be one pixel.
The second project consisted of 12 photographs (11 stereo pairs). The same photographs as in the first project were used. However, not all the photographs were used- the first, third, fifth, etc photographs were used. In this case, fore-and-aft overlap was 82−84%. The project identified 305 tie points with a minimum correlation coefficient of 0,96. According to equalizing results, the mean square error on tie points was 1,273 pixels, i. e., 20 cm on ground.
In creating the third project, we selected the first, fourth, seventh, etc. photographs. As a result, the project consisted of 7 stereo pairs. Fore-and-aft overlap in this
Table 1. — Mean square er
case was 77−79%. During processing, 170 tie points with a minimum correlation coefficient of 0,96 were found. In the previous two projects (23 stereo pairs and 11 stereo pairs), tie points were set manually, as it was difficulty to put them in automatic mode. This is perhaps due to the large fore-and-aft overlap of photographs because excess information interfered. In this project, a third of the total number of points was measured automatically (54 points), while the rest were measured manually. The mean square error (0,399 pixels) for point position finding was calculated — 6.4 cm on ground.
The last project consisted of 5 stereo pairs with a 70−72% fore-and-aft overlap. 139 tie points were found in the project, half of which was identified automatically. The minimum correlation coefficient in this project was 0,96. The mean square error was 0,089, which is less than 1 cm on ground. It is only in this proj ect that all the error values in the report were within permissible values (0,2).
Table 1 shows the processing results for 4 projects in the same area with different fore-and-aft overlaps. s for point position-finding
Number of Number of tie Fore-and-aft Minimum corre- MSE MSE MSE MSE
stereo pairs points in the project overlap (in%) lation coefficient X Y Z Exy (metre)
23 623 90−92 0,96 1,281 2,243 4,637 2,583
11 305 82−84 0,96 0,414 1,204 1,972 1,273
7 170 77−79 0,96 0,224 0,330 0,716 0,399
5 139 70−72 0,96 0,042 0,079 0,140 0,089
According to the «Instructions for Photogrammetric Works When Creating Digital Topographic Maps And Plans», the fore-and-aft overlap of aerial photographs should be at least 60% [7, 12]. A 90−92% overlap is excessive — can lead to large errors.
Comparing the mean square errors of the four projects, it can be seen that the MSE directly depends on the number of tie points set and on the number of
stereo pairs. The more there are stereo pairs on the same territory, the greater the need to put tie points. The use of excess data therefore increases MSE.
Using the data in Table 1, two histograms showing the dependence of mean square error on the number of points and on the number of stereo pairs in proj ects were constructed (picture 3, 4).
Picture 3. Histogram of dependence of mean square error on number of points in a project
Section 2. Geodesy
number of stereo pairs
Picture 4. Histogram of dependence of mean square error on number of stereo pairs
Analyzing the above histograms, it can be concluded that the MSE value directly depends on the number of tie points found in the projects. The more there are stereo pairs in a project on the same territory, the greater the need to determine the tie points.
Comparison of the four projects created showed that the more the fore-and-aft overlap is (above 80%), the larger the aerial photographs. Such number of photographs increases the volume of photogrammetric
work under which one needs to identify a large number of tie points.
With an increasing number ofpoints, the MSE value will increase, thereby reducing the accuracy of determining the location coordinates of the points and, thus, leading to less accurate orthophotoplans.
The use of unmanned aerial vehiclesobtains highresolution images of a small territory to be used to construct high-precision orthophotoplans.
1. Гулиев Ф. Ф. Оптимальная организация экологического мониторинга на базе группы беспилотных летательных аппаратов. — С. -П.: Научное приборостроение, 2012, T. 22, № 4.
2. Васин К. В., Герасимова С. Г. Использование беспилотных летательных аппаратов — новое слово в прогрессивном земледелии. — М.: Геопрофи, № 5, 2014.
3. Никифоров А. А., Мунимаев В. А. Анализ зарубежных беспилотных летательных аппаратов, применяемых в лесном секторе. — Труды лесоинженерного факультета ПетрГУ 2010.
4. Янников И. М., Фомин П. М., Габричидзе Т. Г., Захаров А. В. Применение беспилотных летательных аппаратов при разведке труднодоступных и масштабных зон чрезвычайных ситуаций. — Вектор науки ТГУ № 3 (21), 2012
5. Митин М. Д., Никольский Д. Б. Современные тенденции развития отрасли беспилотных летательных аппаратов. — М.: GEOMATICS № 4, 2013.
6. Фетисов В. С., Неугодникова Л. М., Адамовский В. В., Красноперов Р. А. Беспилотная авиация: терминология, классификация, современное состояние. — Уфа: ФОТОН, 2014.
7. Инструкция по фотограмметрическим работам при создании цифровых топографических карт и планов. ГКИНП (ГНТА)-02−036−02. — М.: ЦНИИГАиК, 2002.

Заполнить форму текущей работой