Evaluation And Extraction Of Forest Resources Information Base On The Wavelet Image Fusion Method

It is the foundation and key-point of study of extraction of changing information of forest resources to improve the quality of spectral and spatial information of remote sensing images. The ZY-1-02 C remote sensing images of Kaihua Zhejiang in 2012 and 2014 are used, including the panchromatic images taken by HRC camera and multi-spectral images taken by PMS camera, to optimize the fusion methods including Brovey, high-pass filter?HPF?, Gram-Schmidt?GS?, IHS, the principal component transform?PCT?, db3 wavelet transform of decomposition level 4?Db3₄?, and the improved fusion method?NIR_WIHS?. On the basis of the optimization of fusion methods, the extraction method of forest resources is further explored for the ZY-1-02 C satellite data.The results are as follow:?1? The optimization of wavelet basis and wavelet decomposition level: According to the quantity evaluation of dispersion standardization method to synthesize the mean, standard deviation, average gradient, entropy, spatial and spectral correlation coefficient and qualitative evaluation, the db3 wavelet basis with the decomposition level 4 is better, whose dispersion standardization indicator is 0.6798.?2? The improved fusion method?NIR_WIHS?: According to the calculation of dispersion standardization indicators of original multi-spectral images, the NIR band has more information than other two bands, at the same time, the near infrared band?NIR? band is located in higher reflection area of vegetation, compared with other two bands. So an improved fusion method combining with IHS and wavelet transformation use the NIR band of original multi-spectral with the same period panchromatic?HRC? images to do the db3 wavelet transform of decomposition level 4, then replace for the I-component of the original multi-spectral images to do the inverse IHS transform.?3? The optimization of fusion methods: Optimize the fusion methods with dispersion standardization method including Brovey, HPF, GS, IHS, PCT, Db3₄ and NIR_WIHS fusion methods. The dispersion standardization indicator of NIR_WIHS method is 0.6863, following the Db3₄ method whose is 0.5888, others are all less than 0.400. So the NIR_WIHS method is much better than other fusion methods to do the visual interpretation with ZY-1-02 C data.?4? The extraction of forest resources information: After using the NIR_WIHS fusion images to do the visual interpretation and ??? Ta criterion method of extraction of forest resources information, the NIR_WIHS method is further proved better than the original multi-spectral images and other fusion images. On the basis of truth value of visual interpretation the results of ??? Ta criterion method is evaluated. The producer accuracy is 92.31%, the user accuracy is 99.08%, the commission is 8.26%, the omission is 0.92%,when the NDVI of sub-compartment is decreasing; the producer accuracy is 99.42%, the user accuracy is 87.24%, the commission is 0.51%, the omission is 12.76%,when the NDVI of sub-compartment is increasing. The overall accuracy is 93.48%, and Kappa coefficient is 0.87, which indicate ??? Ta criterion method is suitable.?5? During the 2012 to 2014, the specific situation of forest resources of study area are as follow:?1?With the process of urbanization during 2012 to 2014, the reduce area of forest land is 0.21% of the original forest land area. Most of them because of the construction and roads. For the sacrifice of forest land area, in the short term may be helpful to the economy growth, however, in the long run of ecological will make economic development under the influence of damping.?2? The increase area of forest land is 0.0015% of the original forest land area. Compared with the rate of decreasing of forest land, effectively promote the sustainable development strategic goal a long way to go.