Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A Time-Space Efficiency Optimization Method for High-Resolution Remote Sensing Image Classification Algorithms

A remote sensing image and classification algorithm technology, applied in the field of computer image processing and remote sensing, can solve problems such as difficulty, insufficient memory, long processing time of remote sensing images, etc. Effect

Inactive Publication Date: 2019-03-05
SHANDONG UNIV OF SCI & TECH
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

At present, the main frequency of computer CPU and memory capacity increase relatively slowly, so it is difficult to solve the problems of long processing time and insufficient memory of remote sensing images by relying on the improvement of hardware performance

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Time-Space Efficiency Optimization Method for High-Resolution Remote Sensing Image Classification Algorithms
  • A Time-Space Efficiency Optimization Method for High-Resolution Remote Sensing Image Classification Algorithms
  • A Time-Space Efficiency Optimization Method for High-Resolution Remote Sensing Image Classification Algorithms

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0023] The invention discloses a method for optimizing the space-time efficiency of a high-resolution remote sensing image classification algorithm. The specific steps of the optimization algorithm are as follows:

[0024] A: The image data is divided into several pixel blocks, each pixel block is 512*512;

[0025] B: Use the RasterIO function in the GDAL function library to load the pixel set of the image block by block, and utilize the memory-mapped file technology to transfer the pixel data block by block to a new disk file;

[0026] C: Randomly select K pixels as the initial cluster center;

[0027] D: Use the memory-mapped file technology to read the pixel data of the first block from the disk file, and transmit it to the device side;

[0028] E: Calculate the distance between each pixel of the block and each cluster center on the device side, and classify according to the principle of similarity;

[0029] F: Send the classification results back to the host, and use mem...

Embodiment 2

[0036] Experiment purpose and method: In order to evaluate the feasibility and effectiveness of this application, this embodiment takes the K-Means algorithm as an example, using a cropped small image (600*600 pixels) and a whole scene image (6000*6000 pixels) were experimented, and for an image of 600*600 pixels, the time spent on a single iteration of the three strategies of "CPU+Memory", "CUDA+Memory" and "CUDA+MMF" were compared; for a 6000*6000 pixel SPOT-5 The entire scene image, the time spent on a single iteration of the three strategies of "CUDA+Memory", "CUDA+MMF" and "CUDA+Blocking".

[0037] Experimental results: as shown in Table 1:

[0038] Table 1 Execution time of four implementation strategies of K-Means clustering algorithm (unit: ms)

[0039]

[0040] It can be seen that the time spent by the "CUDA+Memory" and "CUDA+MMF" strategies is far less than the time spent by the "CPU+Memory" strategy, only about 3% of the latter, and the time spent by the "CUDA+M...

Embodiment 3

[0043] Experiment purpose and method: In order to evaluate the feasibility and effectiveness of this application, this embodiment takes the K-Means algorithm as an example, and performs "CUDA+ memory", "CUDA+ MMF" and "CUDA+blocking" three strategies to compare and evaluate the memory usage.

[0044] Experimental results:

[0045] As shown in Table 2, for a small image of 600*600 pixels, the memory occupied by the "CUDA+MMF" and "CUDA+Blocking" strategies is slightly smaller than the memory occupied by the "CUDA+Memory" strategy; for a medium-sized image of 3000*3000 pixels , the memory occupied by the "CUDA+MMF" strategy and the "CUDA+blocking" strategy is significantly smaller than the memory occupied by the "CUDA+Memory" strategy, only about 25% of the latter; for a large image of 6000*6000 pixels, "CUDA+MMF" And the memory occupied by the "CUDA+blocking" strategy is much smaller than the memory occupied by the "CUDA+memory" strategy, only about 10% of the latter. It can ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to the field of computer image processing and remote sensing technology, and discloses a method for optimizing the space-time efficiency of a high-resolution remote sensing image classification algorithm, which adopts a strategy for block processing of images, and introduces CUDA parallel computing and memory mapping file technology. Avoiding the read and write operations of the disk greatly saves the running time of the algorithm. The rationally optimized algorithm retains the advantages of less memory usage in the block processing method, and improves the image file access speed, improving the time efficiency and time efficiency of remote sensing image processing. space efficiency.

Description

technical field [0001] The invention relates to the technical fields of computer image processing and remote sensing, in particular to a method for optimizing the space-time efficiency of a high-resolution remote sensing image classification algorithm. Background technique [0002] With the rapid development of satellite remote sensing technology, the data volume of remote sensing images is increasing. For example, the SPOT-5 satellite image stores 5.76*108 pixels, and the WorldView-2 satellite image stores 2.62*108 pixels. Such a large amount of data brings great pressure and challenges to remote sensing data processing, mainly in the long running time of the program and insufficient computer memory. At present, the main frequency of computer CPU and memory capacity increase relatively slowly, so it is difficult to solve the problems of long processing time and insufficient memory of remote sensing images by relying on the improvement of hardware performance. How to impro...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62
CPCG06F18/23211G06F18/24137
Inventor 崔宾阁马秀丹赵法喜
Owner SHANDONG UNIV OF SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com