Study Of The Forest Area Statistic Based On The RS Technology

Accompany with the coming of the space times and information times, remote sensing technology develops quickly, and it is also used to acquire the information which can not be acquired from the earth surface. Remote sensing is suitable for forestry with the resource investigation and monitoring. The development of the remote sensing technology supplies a new method for monitoring the changing of the forest resource in time.This paper adopts RS software ERDAS IMAGINE 8.7, studies the forest area statistic based on the image processing. This method can not only avoid some disadvantages of the traditional investigation, such as heavy job, long investigation time, slow speed, and low precision and so on, but also has many merits such as large information and advanced detection method, which can carry on the forest area statistic quickly and precisely. It is also the purpose of this main content studied. Carrying on the forest area and coverage investigation in Pukou Old mountain forestry center for the first time, it will be significant effect to promote the spatial database construction and realize the dynamic monitoring and network management of the forest resource in Nanjing.The main content and conclusion of this research is summed up as follows:(1) The relief map and remote sensing image of the research area have been geometric corrected to the coordinate system of Beijing 54.(2) To improve the visual effect, Image has been enhanced. In this article, concluding color enhance, contrast enhance and ratio enhance. In the color enhance, according to image character, correlation coefficients and so on, choose the band combination and image fusion.(3) To improve the image information and resolution, multi-spectrum image and high resolution image have been fused. Comparing the HIS, PCA, Brovey and wavelet transformation, HIS is the best fusion method.(4) Image classification and Classification accuracy evaluating. Gain the sample by ISODATA unsupervised classification firstly, and then Bayes supervised classification, verify the sample in reality. Computing the error matrix, evaluating the results by User's Accuracy, Producer's Accuracy, Overall Accuracy and Kappa coefficient to evaluate the classification accuracy. The results satisfy the requirement.(6) Compute the area of forest based on the numbers of pixel after classification.The results also indicated that extracting the forest area by remote sensing method precedes the manual investigation; classifying by computer could achieve better accuracy. Except for the computer classifying, we should do some ground survey and validation. As for remote sensing development presently, we should not substitute completely computer auto classification for the ground survey.