Rice mapping method based on self-adaptive feature selection

Abstract

The invention relates to a rice mapping method based on self-adaptive feature selection. The rice mapping method comprises the steps of establishing a time sequence data set of enhanced vegetation indexes and water indexes in a research area; establishing time sequence data of cloud distribution in the research area; based on the cloud distribution of remote sensing images in a rice crucial phonological period, dividing the research area into a cloud area and a cloud-free area; based on a time sequence analysis method, acquiring rice classification results in the cloud-free area; extracting pixel-based remote sensing image features; selecting images with the least cloud interference, and segmenting in sequence to obtain remote sensing image objects respectively for the cloud area and the cloud-free area; integrating the pixel-based remote sensing image features, and extracting object-oriented remote sensing image features; taking the rice classification result in the cloud-free area as training data, and obtaining rice classification result in the cloud area; and integrating the rice classification result in the cloud-free area and the rice classification result in the cloud area, and obtaining a rice spatial distribution map in the research area. The invention has the characteristics of high automation degree, ease of use, good robustness, high classification accuracy and the like.

Description

technical field [0001] The invention relates to a rice mapping method based on adaptive feature selection. Background technique [0002] As rice is the most important food crop in China and even in the world, it is very important to accurately grasp its planting area to ensure food security. At the same time, since the rice growth process requires a large amount of irrigation water and emits considerable methane gas, it is particularly important to quickly and efficiently monitor the spatial and temporal distribution of rice for the sustainable development of regional resources and the environment. The traditional artificial ground survey method is limited by various aspects such as manpower, material resources and natural conditions, and it is difficult to meet the current situational demand for rapid monitoring of crop planting area. With the in-depth development of remote sensing technology, especially the launch of Landsat 8 and China's Gaofen series satellites, which a...

AI Technical Summary

Problems solved by technology

However, it is often difficult to combine the advantages of high temporal resolution with higher spatial resolution data

Images