OPTIMIZATION OF GROUND CONTROL POINT (GCP) AND INDEPENDENT CONTROL POINT (ICP) ON ORTHORECTIFICATION OF HIGH RESOLUTION SATELLITE IMAGERY

. In the rapidly evolving technology era, various survey methods have been widely used one of them by remote sensing using satellite. It is known that the satellite image recording process is covered by rides (satellites) moving over the Earth's surface at hundreds of kilometers, causing satellite imagery to have geometric distortion. To reduce the effect of geometric distortion of objects on the image, geometric correction by orthorectification is done. Pleiades is a satellite of high resolution satellite image producer made by Airbus Defense & Space company. The resulting satellite imagery has a 0.5 meter spatial resolution. As a reference for the more detailed space utilization activities of space utilization arranged in the Regional Spatial Plans, Detailed Spatial Plans was created with the 1: 5000 scale map which has been governed by the Geospatial Information Agency. In the process of orthorectifying satellite imagery for this 1: 5000 scale map, ground control or Ground Control Point (GCP) is used for geometric correction and Digital Elevation Model (DEM) data. In this research, the optimal number of GCP usage for orthorectification process in Rational Function method is 21 GCP using 2nd order polynomial


I. INTRODUCTION
As a reference for more detailed space utilization activities of spatial activities regulated in the Spatial Plan, a Detailed Spatial Plan with a 1: 5.000 scale map has been prepared by the Geospatial Information Agency (BIG). Used on a scale of 1: 5,000, because on a scale map greater than or equal to 1: 5000, in addition to blocks, will also be drawn other elements such as channels and fences. Pleiades is a satellite of high resolution satellite image producer made by Airbus Defense & Space company. The Pleiades satellite produces satellite imagery data in two modes, namely the panchromatic mode and the multispectral mode. Satellite images in the panchromatic mode have a spatial resolution of 0.5 meters with a band number of 1 band (panchromatic), while satellite images in multispectral mode have a spatial resolution of 2 meters with a band number of 4 bands (VNIR -Visible Near Infra Red). This satellite image has good specifications, so it is often used in survey planning and making thematic maps [4]. This satellite image is great for 1: 5.000 scale map with the required horizontal precision of 0.5-2.5m [1]. In the process of orthorectifying this satellite image, a ground control point or Ground Control Point (GCP) is used for geometric correction of x and y positions. This GCP has an important role to correct data and improve the overall image. The accuracy of ground control points depends on the type of GPS used and the number of point samples on the location and the time taken. For 1: 5,000 scale mapmaking, a static geodetic GPS receiver of static method with 30-45 minute observation [1] was used. For the addition of z position information, Digital Elevation Model (DEM) data was added. DEM was a digital data describing the geometry of the earth's surface shape or parts thereof which comprises the set of coordinate points of the sampling result from the surface of the earth with the algorithm defining the surface using the coordinate set. For 1: 5.000 scale mapmaking, DEM can be used with 5-10 meter vertical accuracy [1].
This study aims to examine the effect of the number of GCP with the uniform distribution pattern used in the image orthorectification process to the accuracy of the object analyzed by looking at the change of Independent Check Point position (ICP). It will also find out what the minimum GCP or limit on the process of orthorectification that entered the tolerance according to satellite image resolution and also will be sought GCP limitations so that the addition of GCP in the process of orthorectification does not affect the scale of the resulting image with respect to Root Mean Square Error (RMSE) which generated.

II. DATA AND STUDY AREAS
The location of this research was conducted in Lumajang Regency, East Java. The city coordinates are located at 112 ° 53 '-113 ° 23' East Longitude and 7 ° 54 '-8 ° 23' South Latitude. The research location in Lumajang City is one of the reasons for the orthorectification and not the usual rectification. Due to the relief form that varies from a height of 0-2000 meters. The data used in this research, among others : a. Satellite images of Pleiades 1-A scene 033 with 0.5x meter panchromatic accuracy and 2 meter multispectral keteitian in East Lumajang District. b. DEM SRTM / X-SAR data from TerraSAR-X satellite with a relative accuracy of 6m and an absolute 16 meters lumajang district c. Coordinate data of ground point amount 74 point measured by static methods for 30-45 minutes.

III. METHODS
In general, this research is conducted in five stages of research, namely the preparation stage consisting of problem identification process, literature study and data collection process, data processing stage, analysis phase and final stage in the form of report preparation.
Here is an explanation of the data processing stage: First about data collection. The data required in this research is, satellite image data of Pleiades  (1) where there are 7 parameters, including 4 in the numerator, and 3 and 1 constants in the denominator. In order to solve the Rational Function coefficient in this order of 3, then there are at least 7 control points (GCP) required. The following polynomial orde 2 [3,7] equation is (2) where there are 19 parameters, including 10 in the numerator, and 9 and 1 constants in the denominator. In order to solve the Rational Function coefficients in this 2nd order, then there are at least 19 control points (GCP) required. The following polynomial orde 3 [3,7]

IV.1. RMSE Results Orthorectification
The Placement of GCP Point were located to represent terrain condition in order to increase the accuracy of RMSe. The process of orthorectification using the Rational Function method has 3 choices of polynomial order.

IV.2. Analysis of Distance Accuracy and Test of T-Test Statistics
In this approach, the distance analysis will only be performed on the best model of each order, ie at 1 st order with the use of 9 GCP, 2 nd Order with the use of 21 GCP, 3 rd order with the use of 41 GCP. Here is a sample test of distance from west of image (point 033_104) to east position of image (point 033_69) at α = 5% and α = 10%  It can be concluded that although the best model for 1: 5000 scale map is available on 3 rd order with 41 GCP usage, but for optimal quantity GCP use 2 nd order with 21 GCP.

IV.3. Pixel Image Analysis Result Ortortification
In pixel image analysis ortorektifkasi results done on 2 nd order with the use of GCP as much as 21 point order 2, which in the previous explanation found that the 2 nd order with polynomial order 2 is the optimal model in the scale map 1: 5000. The following changes pixel size after the orthorectification process obtained from 5 sample point.

V. CONCLUSION
The conclusion from this research are : a. In orthorectification process, so that RMSE GCP value ≤ 2.5m can be used amount of GCP as much as 8 GCP with polynomial of orde b. In the process of orthorectification the addition of excessive amount of GCP does not alter the scale of the resulting map significantly. For order 1 orthorectification process, optimal GCP used is 9 GCP, 2nd order is 21 GCP, and order 3 is 41 GCP c. Relation It is found that the order of polynomials used with the resulting image scale is in order 1 can be used for 1: 3000 -1: 11000 scale map creation, order 2 can be used for 1: 1500 -1: 5000 scale map creation, order 3 can be used to map scale ≤ 1: 1500. The scale varies depending on the accuracy class of the map used d. Although the best model in geometric accuracy is found in 3 rd order with the use of GCP as much as 41 points and using the order of polynomial 3. But for the scale map 1: 5000, simply use model 7 with the use of 21 GCP with the order polynomial 2.