Remote Sensing Image Correction and Error Source Exploration —— Taking Jixi City as an Example

(Heilongjiang Provincial Institute of Land and Resources Survey and Planning, Harbin, 150056)

The coordinate accuracy of satellite images is very important for updating the land use database. According to the remote sensing data source of Jixi city, ERDAS IMAGINE is used to correct and analyze the error of the remote sensing image of Jixi city, and the causes of internal and external errors of satellite remote sensing image are discussed.

Keywords: satellite image; Remote sensing; Mistakes; Orthography; Democrat (same as Democrat)

1 Present situation and application prospect of high-resolution satellite remote sensing technology

High-resolution satellite images usually refer to remote sensing images with pixel spatial resolution within10 m. The development and application of early high-resolution sensors were mainly in the military field, aiming at making large-scale remote sensing maps, analyzing ground objects and monitoring human activities. After 1990s, it gradually entered the commercial and civil fields and developed rapidly.

China has developed and launched six meteorological satellites, including sun and geosynchronous orbit satellites. 1999 10 The successful launch of China's first China-Brazil earth resources satellite with land resources and environment as its main observation targets ended the history of China's lack of high-spatial resolution transmission resources satellites. At the same time, China has also established a number of national remote sensing application institutions, which provide a wide range of remote sensing services such as meteorological forecasting, land survey, crop yield estimation, forest survey, geological prospecting, ocean forecasting, environmental protection, disaster monitoring, urban planning, and map mapping.

The appearance of high-resolution satellite remote sensing images makes it possible to observe the detailed changes of the ground surface on a small spatial scale, carry out large-scale remote sensing mapping and monitor the impact of human activities on the environment, which has broad application prospects. The close combination of high spatial resolution image data and geographic information system has been proved to have great application potential in urban ecological environment evaluation, topographic map update, cadastral survey, precision agriculture and so on.

2 remote sensing image processing flow

Remote sensing images generally go through the following steps from reception to final application (Figure 1):

3 Jixi City image data source and remote sensing image deformation reasons

3. 1 Jixi image data source

The image data sources of Jixi city are SOPT-5 satellite data, A 1 level data and * * * two scenes.

Figure 1 Remote Sensing Image Processing Flow

Track number: 305-258; 307-258

Imaging time: 2005/11/22; 2005/ 1 1/ 12

Incident angle: r11.52; R 18.28

CNES, with the participation of Belgium and Sweden, has developed the SPOT Earth observation satellite system. The resolution of SPOT-5 is 2.5m, and the imaging mode is CCD array push-scan imaging. See table 1 for specific parameters.

Table 1 SPOT-5 satellite parameters

SPOT image products * * * are divided into five levels:

Grade 1A: the image is only corrected by radiation, but not by geometry.

1B level: based on 1A level, some geometric corrections are made to correct the deformation caused by panoramic deformation, earth rotation, curvature and orbital height change.

Level 2A: Standard map projection has been added.

Class 2B: geographical correction, with geodetic control points and average height correction added.

Orthographic leveling: complete geographical correction, digital elevation model processing, eliminating the projection error caused by terrain fluctuation.

3.2 Analysis of deformation causes and error sources of remote sensing images

The overall deformation of remote sensing images is the result of translation, scaling, rotation, deflection, bending and other deformations. After receiving the remote sensing data, the receiving department first corrects the data according to various parameters of the remote sensing platform, the earth itself and the sensor, and the provided images have different levels according to different levels of preprocessing.

3.2. 1 the influence of the change of the position and motion state of the remote sensing platform.

Whether it is an airplane or a satellite, the flight attitude will change for various reasons during the movement, which will cause image distortion.

(1) Yaw means that the remote sensing platform deflects by a small angle relative to the original course in the process of advancing, which causes the change of scanning line direction and the image tilt distortion, as shown in Figure 2 (a).

(2) The elliptical orbit of the speed satellite itself leads to the non-uniformity of the satellite speed. When the speed is fast, the scanning belt is ahead; When the speed is slow, the scanning band lags behind. This may cause the position of the image to move in the forward direction of the satellite (up and down direction of the image), as shown in fig. 2 (b).

(3) The altitude is affected by mechanical factors during the movement of the platform, which leads to deviation from the original standard altitude, or the orbit of the satellite itself is elliptical. The flying height always changes, but the scanning field angle of the sensor remains unchanged, which leads to the change of the ground length corresponding to the image scanning line. The higher the altitude deviation, the wider the ground corresponding to the image, as shown in Figure 2 (c).

(4) The elevation change of the elevation remote sensing platform can cause the change of the up-and-down direction of the image, that is, the point below the satellite moves backward when descending, and moves forward when ascending, resulting in the dislocation between lines, as shown in Figure 2 (d).

(5) The attitude rolling of the rolling remote sensing platform refers to rotating an angle with the forward direction as the axis, which can make the under-satellite point shift in the scanning line direction, so that the rolling angle in the line direction of the whole image causes the shift direction to move, as shown in Figure 2 (e).

Fig. 2 the influence of the change of the position and motion state of the remote sensing platform on the image.

3.2.2 Influence of topographic relief

When the topography fluctuates, local image points will be displaced, so that the image points that should have been ground points are replaced by the image points at the same high point.

3.2.3 The influence of the curvature of the earth's surface

The earth is an ellipsoid, so its surface is curved. The surface mainly affects two aspects, one is the movement of pixel position, and the other is the unequal width of pixel relative to the ground. When the scanning angle of the sensor, that is, the incident angle, is large, the influence is more prominent, resulting in the edge scene being compressed when the image is displayed. Assuming that the real scene on the original ground is a straight line, the center is narrow and the edge is wide when imaging, and the pixel size is the same when displaying a single image, then this straight line is displayed as anti-S-shaped bending, which is also called panoramic distortion.

3.2.4 Influence of atmospheric refraction

The influence of atmospheric refraction is that the atmosphere refracts the propagation of radiation. Because the density distribution of the atmosphere is getting smaller and smaller from bottom to top, and the refractive index is constantly changing, the propagation of refracted radiation is no longer a straight line but a curve, which leads to the displacement of the image point received by the sensor.

3.2.5 Influence of the Earth's Rotation

When the satellite is running, when the sensor scans the ground to obtain an image, the image will be biased due to the rotation of the earth. Because most satellites receive images during the orbital descent phase, that is, the satellites move from north to south, when the earth rotates from west to east, the satellite sub-star points gradually deviate.

Through the above deformation analysis of remote sensing images, we can see that the errors can be basically divided into three categories: the internal errors caused by the remote sensor itself vary with the structure, characteristics and working mode of the remote sensor, and such errors are generally small; External errors caused by external factors, such as deformation analysis of remote sensing images; The machining errors produced in the machining process are human errors.

Orthophoto correction of Jixi remote sensing image

4. 1 natural conditions and data of Jixi city

4. 1. 1 natural conditions

Jixi City is located in the southeast of Heilongjiang Province and is known as the "Spring City in the North". It borders linkou county and Muling County of Mudanjiang City in the west, Raohe County, Baoqing County, qitaihe city County and Boli County of Shuangyashan City in the north, Wusuli River and Songacha River in the east, and Xingkai Lake, Bailing River and land in the south. It is a border city in the eastern part of the province, with a length of 687.5km. There are six districts, two counties and one city in the administrative area of Jixi City, with a total land area of 22488.46km2. Jixi City is rich in mineral resources, dotted with tourist attractions, rich in special tourism resources such as ice and snow, lakes, wetlands and forest tombs, and the terrain is complex.

4. 1.2 data

Jixi city uses twice spot-5 data and 14 1∶50000 DEM, covering all six districts of Jixi city.

The original DEM is 1∶50000 digital elevation model, grid format, vertical interval 10 m,1980xian coordinate system, 1985 national elevation datum, Gauss-Kruger projection, and 6-degree zoning.

Control points are measured by GPS, and 80 GPS points are measured by two images.

4.2 Jixi image correction workflow

4.2. 1 control point selection and DEM data processing

Each scene image selects 40 control points, and geodetic coordinates are measured by field GPS. Due to the limitation of other environmental factors, 36 points were measured according to the original puncture points, and the remaining 4 points were measured by field surveyors according to the actual situation, and marked on the base map to record the detailed points.

The GPS model is Ling Rui S80；; The accuracy is 5 mm+ 1 ppm.

Fig. 3 Workflow of image correction

Original DEM *** 14, standard format of State Bureau of Surveying and Mapping, 1∶50000 digital elevation model, grid format, vertical interval 10 m, 1980 Xi 'an coordinate system, 1985 national elevation datum, Gauss-.

The final image needs to be corrected to 1954 Beijing coordinate system, 3 degree zone, so DEM also needs to be corrected to this coordinate system.

Convert the format first, then embed it into the file, and finally convert the coordinate system from 1954 to 1980 and project it into the 3-degree area.

The processed DEM is a digital elevation model of 1∶50000, IMG format, vertical spacing 10 M, Beijing coordinate system 1954, Gauss-Kruger projection, 3-degree zone. All DEM data are embedded in IMG file, covering the whole monitoring area.

4.2.2 Correction of panchromatic images

ERDAS 8.7 is used as the operating software to read the DIM file of the original panchromatic image, add the satellite parameters into the calibration model, and then add the processed DEM data. At this time, the panchromatic image data already has approximate geographical coordinates, and the error is still very large. According to the feature points on the image, input the coordinates of the GPS points measured on the spot, and fine-tune the image feature points corresponding to the GPS points to make their positions closer to the real coordinates and reduce the control point error. Analyze the causes of the points with large errors, adjust and delete them according to the specific situation, and correct the image after meeting the correction conditions. The resolution of image resampling is 2.5m, and the resampling method is cubic convolution processing. The coordinate distribution of image control points and their correction errors are shown in Table 2.

In the application of high resolution image correction, the selection of data source is very important. Different satellites, different images and satellite images of different quality have great influence on the correction method and final correction accuracy. DEM data is very important for mountainous areas. Good satellite images, accurate coordinates of feature points and suitable DEM data can minimize external error factors in satellite image correction.

Table 2 Statistical table of residual error of control points of 307 -258 scene corrected images in Jixi City

refer to

Hu mingcheng Development and Latest Achievements of Satellite Remote Sensing Technology [J]. Surveying and Mapping Science, 2000,25 (1)

Qiu,,,. Remote sensing application technology [M]. Wuhan: Wuhan Technical University of Surveying and Mapping Press 1995

Sun jiabing Principle and Application of Remote Sensing [M], Wuhan: Wuhan University Press. 2003

Mei Anxin and others. Introduction to Remote Sensing [M]. Beijing: Higher Education Press. 2003

Party, Wang Xiaodong, et al. Remote sensing image processing method of ERDAS IMAGINE [M]. Beijing: Tsinghua University Publishing House, 2003.

Ch/t 1008-200 1。 Digital elevation model of basic geographic information products 1: 10000, 1: 50000 [s]. Beijing: China Standards Publishing House, 200 1.