Study On Aspect Extraction Of Forestry Subcompartments Based On DEM

Today “Digital Forestry” continues to thrive,the connection between forestry science and geographic information science becomes more and more strong.With the development of digital terrain analysis technology,to extract region aspect of subcompartments automatically from DEM is especially common during forest resources survey.Compared with the traditional investigation method,it saves lots cost both on labor and time,conforms to the development of forestry informatization.However,the exist algorithms of subcompartment region aspect extraction still have different defects and how to improve the accuracy and reliability is an urgent problem.Therefore,on the basis of previous studies and to fit the fundamental demand of high-precision data of forestry survey,this paper proposes a new subcompartment region aspect extract algorithm,which considers reasonable mathematical models and scientific theoretical basis.In this paper,we first introduce the general situation of this study area and the study data needed,also make the corresponding pretreatment of the study data so that it can meet the basic requirements of the experiment.Then,the algorithm of single cell aspect extraction is studied.It is found that the third-order finite difference algorithm has the best accuracy in the real DEM.Therefore,it is determined to be the study method of the single cell aspect extraction in this paper.In addition,we find out the principle of existing algorithms about subcompartment aspect extraction and do the error analysis.Concluded that both Vector method and Maximum characteristic method get error from single cell aspect,while the previous one contains statistical error and weight error and the next one contains truncation error and mode error.After understanding the source and property of error,this paper put forward two new algorithms named Main Direction Weighted Vector and Fuzzy Pattern Recognition.The Main Direction Weighted Vector algorithm uses the modulus of each cell as the weight in algebraic statistics,determines the main direction of statistical quantitative results,rejects the interference data,and finally,obtains the region aspect grade by qualitative description.The Fuzzy Pattern Recognition algorithm constructs the subcompartment fuzzy matrix and template matrix through the membership function,calculates the closeness degree,and finally,determines the region aspect grade according to the near principle.We use C# advanced programming language to realize algorithm automation in this paper,and the experimental results show that the Main Direction Weighted Vector and the Fuzzy Pattern Recognitionhave similar distribution in the different classification groups.It is proved that the tow algorithms are similar to the result of subcompartments aspect extraction.In addition,the two proposed algorithms are feasible and accurate comparing to the existing ones.Futhermore,due to the adaptability experiment we find that the Main Direction Weighted Vector algorithm is more suitable for the circular pattern spot,while the Fuzzy Pattern Recognition algorithm is more suitable for the strip pattern spots.Finally,two algorithms have no significant difference when facing different subcompartments within different terrain factors.