Scene classification in very high resolution (VHR) remote sensing (RS) images is a challenging task due to complex and diverse content of the images. Recently, convolution neural networks (CNNs) have been utilized to tackle this task. However, CNNs cannot fully meet the needs of scene classification due to clutters and small objects in VHR images. To handle these challenges, this letter presents a novel multi-level feature fusion network with adaptive channel dimensionality reduction for RS scene classification. Specifically, an adaptive method is designed for channel dimensionality reduction of high dimensional features. Then, a multi-level feature fusion module is introduced to fuse the features in an efficient way. Experiments on three widely used data sets show that our model outperforms several state-of-the-art methods in terms of both accuracy and stability.

No abstract

No abstract

No abstract

Image fusion is of great importance to various remote sensing applications because many Earth observation satellites provide both high-resolution panchromatic (Pan) and lowresolution multispectral (MS) images. A number of fusion methods have been proposed, such as intensity-hue-saturation fusion and wavelet transform fusion methods. However, further studies are still necessary to improve the fusion performance for new types of remotely sensed images, such as IKONOS or QuickBird images. We propose an improved bilateral total variation filter method for fusing such MS and Pan images based on regularization. First, the constraints on the MS and Pan images are imposed based on the observation model. Then, the improved bilateral filter is used as an a priori model to constrain the high-resolution MS images. Finally, the steepest descent optimization algorithm is used to obtain the estimated MS images. Fusion simulations on spatially degraded IKONOS and QuickBird images, whose original MS images are available for reference, respectively, show that the proposed approach has better spatial quality while keeping the spectral information of the MS images. C 2011 Society of Photo-Optical Instrumentation Engineers (SPIE).