TY - GEN
T1 - Analysis of high resolution panchromatic satellite images, based on GPGPU programming
AU - Aiquipa, Wilver Auccahuasi
AU - del Carpio, Jorge
AU - Garcia, Jorge
AU - Benites, Raul
AU - Grados, Juan
AU - Flores, Edward
N1 - Publisher Copyright:
© 2019 Association for Computing Machinery.
PY - 2019/10/8
Y1 - 2019/10/8
N2 - Satellite images are widely used in the study of land cover, being used for many tasks such as: the evaluation of urban growth, the monitoring of cultivation areas, the assessment of natural disasters and other applications, to perform these tasks is resorted to to the use of satellite images, they provide a variety of information that depends on the optical instrument carried on board in their payload, the information in the observation satellites, is provided through a data matrix that can be represented by a image, one of the characteristics is the spatial resolution where a pixel in the image corresponds to a coverage in meters on land, with this spatial resolution we can evaluate large tracts of land, among the spatial resolutions, we have the metrics where a pixel in the image corresponds to more than one meter on the ground and the sub meter, where a pixel in the image corresponds e less than one meter on land. With the sub-metric resolution we have a greater detail of the area of interest on land, having the disadvantage that a smaller area of land is evaluated in comparison with the metric resolution. The image provided by the satellite is composed of a set of matrices commonly called spectral bands, characterized by the bands of colors: red, green, blue, we also have the middle and near infrared bands, among others; additionally, the band of the panchromatic is presented, which is the band where the maximum spatial resolution is identified, therefore the image size is of high resolution. A methodology is presented to process the panchromatic satellite images through the use of the GPGPU programming using the MATLAB tool, a test with a high resolution image and with a weight in 1270 Megabits, with a size of 26012 X 25512 pixels, was performed. which was applied an algorithm where it evaluates the value of the pixel analyzing the whole matrix of the image pixel by pixel, the calculation was made in a Core i7 CPU with a processing time of 2.81 hours, with GPGPU programming using a card GTX1050Ti graphics was processed in a time of 1.65 hours, achieving the same result in a shorter time compared to the CPU processing time.
AB - Satellite images are widely used in the study of land cover, being used for many tasks such as: the evaluation of urban growth, the monitoring of cultivation areas, the assessment of natural disasters and other applications, to perform these tasks is resorted to to the use of satellite images, they provide a variety of information that depends on the optical instrument carried on board in their payload, the information in the observation satellites, is provided through a data matrix that can be represented by a image, one of the characteristics is the spatial resolution where a pixel in the image corresponds to a coverage in meters on land, with this spatial resolution we can evaluate large tracts of land, among the spatial resolutions, we have the metrics where a pixel in the image corresponds to more than one meter on the ground and the sub meter, where a pixel in the image corresponds e less than one meter on land. With the sub-metric resolution we have a greater detail of the area of interest on land, having the disadvantage that a smaller area of land is evaluated in comparison with the metric resolution. The image provided by the satellite is composed of a set of matrices commonly called spectral bands, characterized by the bands of colors: red, green, blue, we also have the middle and near infrared bands, among others; additionally, the band of the panchromatic is presented, which is the band where the maximum spatial resolution is identified, therefore the image size is of high resolution. A methodology is presented to process the panchromatic satellite images through the use of the GPGPU programming using the MATLAB tool, a test with a high resolution image and with a weight in 1270 Megabits, with a size of 26012 X 25512 pixels, was performed. which was applied an algorithm where it evaluates the value of the pixel analyzing the whole matrix of the image pixel by pixel, the calculation was made in a Core i7 CPU with a processing time of 2.81 hours, with GPGPU programming using a card GTX1050Ti graphics was processed in a time of 1.65 hours, achieving the same result in a shorter time compared to the CPU processing time.
KW - CPU
KW - GPU
KW - Panchromatic
KW - Processing
KW - Satellite image
UR - http://www.scopus.com/inward/record.url?scp=85077967412&partnerID=8YFLogxK
U2 - 10.1145/3365245.3365253
DO - 10.1145/3365245.3365253
M3 - Contribución a la conferencia
AN - SCOPUS:85077967412
T3 - ACM International Conference Proceeding Series
SP - 45
EP - 48
BT - SSIP 2019 - 2019 2nd International Conference on Sensors, Signal and Image Processing
PB - Association for Computing Machinery
T2 - 2nd International Conference on Sensors, Signal and Image Processing, SSIP 2019
Y2 - 8 October 2019 through 10 October 2019
ER -