The Research On Improving The Effective Extraction Of Cultivated Land Information By Using Multi-spectral And Panchromatic Remote Sensing Images

Cultivated land resource is the foundation of social stability and economic development.The modern remote sensing image processing technology can realize the lean management of cultivated land resources.This paper contains the cultivated land information of multi-source remote sensing image data,through the image fusion algorithm and method of research,improvement and comparative analysis,implements the effective extraction of cultivated land in the remote sensing image information.Fusion algorithm using wavelet transform combined with IHS transform,the multispectral image of I component and panchromatic images of high and low frequency component with"bior3.7" wavelet base layer 2 decomposition reconstruction,that integrated the technical advantages of two kinds of algorithm.The experimental results and evaluation show that the fused images with good spatial resolution and spectral information on the average gradient,peak signal-to-noise ratio and so on five indicators.That is superior to the fusion results such as the IHS transform,PCA transform and Gram-Schmidt transformation method.The reference indexes is the spectral characteristics of cultivated land on the red band(mean 503.83,standard deviation 26.48,characteristic value 1901.71)and the NDVI index(0.05092?0.07791).Combine with these object properties in ENVI software based on watershed algorithm Extract Segment module combination optimization method,and the experimental results show that the method not only avoid the Extract Segment module the defect of the optimal segmentation threshold is difficult to select merger,but also solve the problem of the "split" effectively,incorporating similar features reasonable,heterogeneous features accurate,preliminary Extract experiment in cultivated land area of 2.09KM2 and 1.29KM2.Wavelet transform combined with IHS transform fusion algorithm not only successfully applied in farmland information of remote sensing image fusion,also for the extraction of effective cultivated land information provides the high quality data sources.Combined with the attributes of the object segmentation method is contributed to the arable land information extraction effectively,provide reference for Ningxia ecological environment features target classification and recognition.