Remote sensing image forest land change detection method based on sparse DBN model

Abstract

The invention discloses a remote sensing image forest land change detection method based on a sparse DBN model. The method comprises: step 1, segmenting a front-stage high-resolution remote sensing image and a rear-stage high-resolution remote sensing image; step 2, calculating vegetation coverage indexes of all image spots; step 3, adding sparse regular terms into the basic unit restricted Boltzmann machine of the deep belief network, and stacking the sparse restricted Boltzmann machine to form a sparse deep belief network; step 4, training a sparse deep belief network by using the sample obtained in the step 2 to obtain a sparse deep belief network model for forest land detection of the high-resolution remote sensing image; and step 5, automatically classifying all image spots by using the trained sparse depth confidence network. A sparse coding method is adopted on the basis of a human visual information system, sparse constraint is introduced into the DBN to obtain a sparse DBN model, and then high-resolution remote sensing image change detection is carried out on the basis of the sparse DBN model to obtain a forest land change detection result.

Description

technical field [0001] The invention belongs to the field of remote sensing image data processing, and relates to a method for autonomously learning images based on a sparse deep belief network (DBN) model, further automatically classifying images, and finally extracting forest land change information from high-resolution remote sensing images. Background technique [0002] With the implementation of the national sustainable development strategy, marked by the full start of the six key forestry projects, my country's forestry has entered a new stage of comprehensively promoting leapfrog development guided by the theory of sustainable development. However, the current forest resource survey and monitoring is mainly based on manual surveys on the ground. The technical means are backward, and the informatization and automation level of monitoring data collection, transmission and storage are low. It is greatly affected by human factors, and it is necessary to promote the develo...

AI Technical Summary

Problems solved by technology

[0004] Forest change detection is one of the necessary tasks for annual forest resource investigation and supervision. Conventional high-resolution remote sensing image forest change detection is mostly based on pixel change detection, ignoring the large and rich details of high-resolution remote sensing images.

Images