Unlock instant, AI-driven research and patent intelligence for your innovation.

A technology based on probabilistic position model to eliminate the influence of spatial position error on the accuracy evaluation of remote sensing soft classification

A technology of classification accuracy and spatial location, applied in image analysis, character and pattern recognition, image enhancement, etc., can solve the problem of low weight, achieve good elimination effect, eliminate uncertainty, and improve practicality

Active Publication Date: 2021-04-16
CHANGZHOU UNIV
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The defects of this model are: (1) the exponential function is used as the distance attenuation function, and the weight is too low when the spatial position error is greater than 1 pixel; (2) the spatial position error is not random, and has directionality in spatial distribution. This is not reflected in the distance decay function

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 technology based on probabilistic position model to eliminate the influence of spatial position error on the accuracy evaluation of remote sensing soft classification
  • A technology based on probabilistic position model to eliminate the influence of spatial position error on the accuracy evaluation of remote sensing soft classification
  • A technology based on probabilistic position model to eliminate the influence of spatial position error on the accuracy evaluation of remote sensing soft classification

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0054] A technology based on a probabilistic position model to eliminate the influence of spatial position errors on the evaluation of remote sensing soft classification accuracy, comprising the following steps:

[0055] Step 1): Geometric registration of remote sensing images: Correct a pair of remote sensing images (assuming the total number of pixels is N) using the geometric fine correction algorithm to obtain the corrected root mean square error RMSE and the average direction of error offset during the geometric registration process .

[0056] Step 2): Soft classification of remote sensing images: use a soft classification algorithm (linear decomposition algorithm or SVM soft classification method) for remote sensing images after geometric registration to obtain remote sensing soft classification images, and set the type vector at (x, y) on the image is υ(x,y);

[0057] Step 3): Select a reference image for soft classification accuracy evaluation: select a reference imag...

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 belongs to the field of remote sensing uncertainty, and relates to a technology based on a probabilistic position model to eliminate the influence of spatial position errors on the evaluation of remote sensing soft classification accuracy, comprising the following steps: step 1): geometric registration of remote sensing images; step 2): remote sensing images Soft classification; Step 3): Select a reference image for soft classification accuracy evaluation; Step 4): Eliminate the influence of spatial position errors. Compared with the current fuzzy location model, which is the best at eliminating the influence of spatial location errors, the fuzzy location model can eliminate 13% of the kappa-error at most. However, the probabilistic location model proposed in this patent can net eliminate the 16.5% to 35.2% kappa-error, which proves that the probabilistic location model of this patent is better than the fuzzy location model.

Description

technical field [0001] The invention belongs to the field of remote sensing uncertainty, in particular to the influence of space positioning (GPS) error and geometric registration error on remote sensing image soft classification and analysis based on soft classification and a method for eliminating the negative influence. Background technique [0002] Aiming at the transmission and influence of spatial position error in remote sensing analysis, there are currently three methods to eliminate the impact of spatial position error. [0003] (1) Coarse evaluation unit. The main idea of ​​this method is to convert from pixel as evaluation unit to block or polygon as evaluation unit for accuracy evaluation. However, the effect of this method is not obvious in the actual use process. E.g. Stehman believes that although blocks and polygons can eliminate the impact of spatial position errors to a certain extent, the impact on the overall accuracy (OA) and kappa coefficient is stil...

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): G06K9/62G06T7/30
CPCG06T7/30G06T2207/10032G06F18/217G06F18/241
Inventor 顾建宇谭园园卢佩昀徐沁岩
Owner CHANGZHOU UNIV