基于LW一CBAM的荒漠草原植被蓋度提取方法研究

中圖分類號(hào)：S812；TP751 文獻(xiàn)標(biāo)識(shí)碼：A 文章編號(hào)：2095-5553（2025）07-0111-07

Abstract：Inorder toextractdesert grassland fractionalvegetationcoverage inrealtime，accuratelyandquickly，this paper proposedalightweightnetwork model method integrated atention mechanism（Lightweightnetwork-Convolutional Block Atention Module，LW—CBAM）based on the collected UAV hyperspectral remote sensing data.This method improved the traditional 2Dconvolution kernel to 3Ddeeplyseparableconvolution kernel，andcombined the multi-branch methodand theatention mechanismmoduletomakethe modellightweightandimprovedtheaccuracyof themodel.Inordertoobtain the optimal model，this paperoptimized the batch size and learning rateof the model.Theresultsshowed thatcompared with popular deep learning methods such as ResNet34，VGG16，MobileNetV2andMobileNetV3，LW—CBAM had a higher classification accuracy，OA was 98.97% ， Kappa coefficient was 97.94，and the model had a higher estimation accuracy for fractional vegetation coverage.The absolute error from the true value was only 0.17% .TheLW— CBAM's parameter count was reduced by over 90% compared to the other models，and its computational requirements were respectively 1.37% ， 0.74% ， 13.33% ，and 14.81% of the four other models. During the model validation stage，the estimation error of fractional vegetation coverage by LW—CBAM was below 0.3% . This model provided a feasible method for estimating fractional vegetation coverage in desert steppe and provided a basis for grassland degradation control.

Keywords：fractional vegetation coverage；hyperspectral remote sensing；deep learning；lightweight network; attention mechanism;desert steppe

0 引言

草原是我國(guó)生態(tài)系統(tǒng)中不可或缺的一部分[1]。我國(guó)的草地面積將近 4×10⁸hm² ，居世界第二位[2]。然而，自從20世紀(jì)60年代以來(lái)，草原退化現(xiàn)象在全國(guó)各地出現(xiàn)上升趨勢(shì)[3]，黨的二十大報(bào)告指出，推行草原森林河流湖泊濕地休養(yǎng)生息，提升生態(tài)系統(tǒng)多樣性、穩(wěn)定性、持續(xù)性。全國(guó)第六次荒漠化和沙化土地監(jiān)測(cè)調(diào)查結(jié)果顯示，內(nèi)蒙古荒漠化土地占全國(guó)荒漠化土地面積的 23.04%^[4] ，土地沙化、次生鹽漬化嚴(yán)重，生態(tài)修復(fù)的任務(wù)艱巨[5]

草原荒漠化主要表現(xiàn)為植被蓋度（FVC）減少、植被種類減少、裸地面積增加等，因此， FVC 是荒漠化監(jiān)測(cè)的重要指標(biāo)[6。傳統(tǒng)的植被蓋度監(jiān)測(cè)方法以地面實(shí)測(cè)為主，優(yōu)點(diǎn)是精度高、數(shù)據(jù)可靠，但只能在小范圍內(nèi)進(jìn)行植被蓋度計(jì)算[7]。得益于技術(shù)的發(fā)展，衛(wèi)星遙感法可以在大范圍內(nèi)對(duì)植被蓋度進(jìn)行反演，但其分辨率不夠高，成本也比較高昂[8]。近年來(lái)，無(wú)人機(jī)（UAV）搭載高光譜儀借助成本低、分辨率高、操作靈活等優(yōu)勢(shì)迅速被大范圍應(yīng)用9，可為荒漠草原植被蓋度的計(jì)算提供技術(shù)支持。

1數(shù)據(jù)采集與預(yù)處理

植被蓋度通常被定義為植被（包括枝、莖、葉）在單位面積內(nèi)的垂直投影面積所占百分比[10]。張燕斌等[11]采用改進(jìn)的3D—ResNet模型對(duì)荒漠草原地物進(jìn)行分類，其總體分類精度為 97.73% ，為荒漠草原整體生態(tài)系統(tǒng)研究奠定基礎(chǔ)； Xu 等[12]提出了MSR—3DCNN模型用于高光譜圖像分類，并在3個(gè)開源數(shù)據(jù)集上進(jìn)行試驗(yàn)。深度可分離卷積（DSC）[13]降低了模型參數(shù)量，為深度學(xué)習(xí)模型的輕量化提供新思路。……