Research On The Ratio Band Of Remote Sensing Image In Forest Stock Volume And Choice Of Texture Information

Forest stock volume is the summation of existing stem volume in a certain area of forest.Usually, the amount of volume statistics is according to the number of tree trunk with bark.They are the important information of people living environment, planning production energyand forestry protection. In our country, the traditional way to obtain the stock volume isartificial field measurement that the data is estimated through the sample area information. Inthat way, it has the low precision, personal dangerousness in somewhere and costs lots of timeand money. With the3S integrated technology, it has focused on the forestry remote sensingresearch that using the information of high resolution remote sensing image to estimate thestock volume so that the efficiency and precision can be improved.In the volume estimation based on remote sensing and GIS, it is related to sample datacollection, remote sensing information extraction, volume model establishing, precisionanalysis and production application. The core of the final estimate equation precision isdetermined by the detection and selection of GIS factors and ratio band related to theestablishing of volume estimating model in remote sensing image. In this paper, after theanalysis of principal component, quantity of information and correlation of the remote sensingfactors, it selects the factors which have large effects and weak correlation to stock volume.The screening methods consist of ridge estimation, the average of residual sum of squarescriteria, Cp criteria and AIC criterion. Then, using the principles of volume estimatedresiduals and forecast deviation to evaluate the selection methods, it builds the variablepreferred selection system.After determining the variable preferred selection system, the texture characteristicparameters of sample area in remote sensing image are extracted through gray levelco-occurrence matrix, and involved in the building of volume estimation model. By testingthe effect of texture information to stock volume, it proves that the texture information means a lot in estimation model and also improves the accuracy of volume estimation.