Study On Remote Sensing Images Of Forest Vegetation Textures Adaptive Filtering

Forest resources are the important material basis for the development of society,the reproducible trait to economic development,social progress and environmental stability is significant,can't be replaced by other resources.With the rapid development of remote sensing technology and the wide application of information such as the texture and shape features is very rich.In recent years,the use of filtering technology to capture image texture characteristic,has become one of the main study direction of improve the image segmentation accuracy.Integrated use of spectral information and texture information of forest vegetation extract image internal texture structure,become the important direction in the field of image research and development.Because of the species diversity of forest vegetation in remote sensing image,the region and the influence of climate led to the division of forest vegetation is difficult.Makes the fixed parameters of texture filtering method can't exactly express the characteristics of texture,affect the segmentation accuracy.Therefore,this paper proposes a spectral histogram statistics under the model of adaptive texture filtering algorithm,the algorithm is based on object feature attribute to set the texture filtering parameters.Under the texture spectrum histogram statistical model,first from typical area of forest vegetation,based on blue noise feature theory to detect the typical area of forest vegetation.The typical area of forest vegetation of gray level co-occurrence matrix texture feature statistics,get the basis attributes of typical forest vegetation texture,mainly including the scale of the forest vegetation texture primitives,direction and the gray level distribution and so on.Spectral histogram feature extracted by combining multiple filtering function,selective enhancement of image texture feature.According to the prior knowledge and typical forest vegetation texture feature selection filter function,and set adaptive filtering parameters,including the filter window size,direction,frequency and intensity,etc.Compared the two methods filtering effect can be clearly see that advantages of the method to extract the object texture more than using fixed parameters to set filtering method.Image segmentation effect depends on the accuracy of the feature extraction stage,therefore,has a lot to improve the method for the final segmentation effect.In addition,the scale of the integral in spectrum histogram statistics window is set to the segmentation result also has a certain influence,this article also proved through the experiment with object segmentation texture primitive as integral scale,the segmentation process as much as possible to avoid the interference of other spatial texture,great significance to improve the segmentation accuracy.Further research work will get method of improving typical forest vegetation,the image sparse forest vegetation and forest vegetation in the area is too small,in order to ensure that the forest vegetation texture characteristics more precisely,by such means as joining together to get the typical area of forest vegetation.