Super-resolution remote sensing data fusion aims to compose the output image with the higher spatial and spectral resolution from the lower resolution input ones captured for the same territory. The images used for fusion are usually multi-temporal. However, existing multi-temporal image fusion methods exploit only cloud-free images that might be difficult to obtain for some territories where the weather conditions are moderately cloudy during the observation period. In this paper, the clouds and their shadows are considered as scene distortions i.e. significant local changes in brightness caused by some opaque objects or their shadows partially overlapping the scene at the moment of image registration. Here, we propose a multi-temporal remote sensing data fusion method adapted to the dataset containing images partially occupied by scene distortions. The method is based on gradient descent optimization procedure with scene distortion elimination in each iteration. The experiments with the modeled data revealed that our method provides spatial and spectral super-resolution even for datasets including images with scene distortions. In comparison with the scenedistortion free case, the proposed method reduces a root mean square error of the resulting image from 2 to 4% on average in the case of the mixed data sets with few undistorted images (from two to six). The overall research has shown that in the case of lack of the data without scene distortions, additional partially distorted images can be used to obtain better fusion results.
Multi-temporal Earth remote sensing images of the same territory may include random scene-distortions coming from the different natural phenomena, for example, clouds or shadows. These distortions are time dependent and may appear only in several images of the analyzed image set. Thus, they define irrelevant image parts that should be eliminated in the further data fusion process. In this article, we suggest an algorithm for detecting such scene distortions using a series of multi-temporal remote sensing images. The algorithm is based on super-pixel segmentation and anomaly detection methods. The algorithm produces a mask of random scene-distortions for each of the images in the analyzed series. The resulting mask could be used to take into account only the scene-relevant parts in the data fusion methods. The proposed approach allows processing images with different spectral and spatial sampling simultaneously that is very useful for multi-sensor data fusion. We tested an algorithm quality by modeling a series of multispectral images with different parameters of spectral and spatial sampling under the different conditions of cloudiness and cloud shadows as an example of random distortions in the scene. As a result, it was shown that the algorithm provides the accuracy of scene distortion detection about 90% and false detection about 10%.
Special forensic examinations of vegetation areas become an important part of the practice preceding to different land rearrangement activities for the purposes of environmental conservation and rational land use. A traditional approach to such examinations includes laborious ground surveys and produces approximate damage estimations for the large sites. Fortunately, remote sensing (RS) data delivers impressive opportunities to simplify the examinations and increase their accuracy. Thus, the substitution of the traditional ground-survey based methods with RS-data oriented technologies is an important problem. In this paper, we propose a method for plant species concentration (PSC) estimation using RS images that can be applied in special forensic examinations of vegetation areas. PSC is one of the key factors indicating vegetation area status. PSC describes a fraction of the land area covered by the plants of particular species. For example, a tree concentration in the agricultural fields indicates the time elapsed from the last processing date and may be useful for abandoned-field determination. The proposed method assumes that the examined area contains a finite number of target vegetation classes. The expert puts several points in the image for each class without exhaustive border delineation to define typical class representatives. Then, the superpixel segmentation and feature extraction algorithm with further kmeans clustering are used to get the entire study-area classification. Finally, PSC is computed as the elementary vegetation class concentration. Keeping in mind abandoned-field determination, we evaluated our method with simulated and real RS images containing four classes: sparse grass, low grass, high grass, and trees. We found that shadows should be defined as a separate class to minimize estimation errors in real images. Moreover, the superpixel segmentation increases the PSC accuracy by 28% with respect to simple per-pixel clustering. Thus, the experimental results proved the applicability of our method for PSC estimation.
Many ecological applications (e.g., human activity management in protected areas, conservation status assignment) require actual forest inventory data. However, frequent field research of forest areas is very expensive. Therefore, in practice, forest inventory data are slowly updated, approximately one time per decade. From the other hand, modern remote sensing systems combine high-quality imagery with short revisiting time and can be used for the forest inventory data clarification. Our paper presents an investigation of tree species classification based on seasonal Sentinel-2 data (2018) and the latest forest inventory information (2013–2014). The main advantages of Sentinel-2 satellites are a short revisiting time and a large field of view that is important in large area analysis. Our classification model was based on support vector machines method combined with specific spatial processing methods. We used the known inventory data for training and validation the classifier. Misclassified regions were further analyzed in ground surveys to produce the inventory data clarification. The paper addresses the optimal image dates selection, image preprocessing and classification procedure evaluation issues. The study was carried out for the territory of the Krasnosamarskoe forestry in Samara region, Russia. The experiments have shown that the proper Sentinel-2 data selection and classification procedure configuration allow reaching the classification accuracy of about 0.82 for the control sample. The ground survey confirmed that classification errors are mainly caused by the dominant tree species changes. Thus, we concluded that Sentinel-2 data can be effectively used for the forest inventory data clarification.
Multi-sensor remote sensing image super-resolution aims to provide better characteristics for different types of resolution and compensate the limitations of the particular imaging systems. However, existing super-resolution techniques consider spectral and spatial resolution enhancement separately, i.e. only spatial or only spectral resolution can be enhanced. Among spatial super-resolution methods maximum a posteriori estimation approach with B-TV regularization stands out as one of the best method for spatial resolution enhancement. But existing implementations were designed only for RGB and grayscale photographic imagery. Unlike photographic RGB imagery, multispectral remote sensing images captured by optical sensors often contain more than three spectral channels (red, green and blue) and, moreover, different remote sensing systems produce a different spectral response for the similar spectral components. Therefore, a more complex image acquisition model should be regarded to take into account the variations in bandwidth and number of spectral channels in the case of remote sensing images. In this article, we propose an algorithm aiming to provide spectral-spatial multi-sensor remote sensing image super-resolution. We apply a joint spectral-spatial image acquisition model, that is typical for remote sensing systems, and investigate the super-resolution algorithm streaming from this model and the maximum a posteriori estimation approach with B-TV regularization. We propose a simple way to adapt B-TV regularization in the case of multiple spectral channels. Our experimental results confirm the enhancement in the spectral and spatial resolution of the output image in comparison with the input images. The results of our research demonstrate that the proposed method achieves both spectral and spatial super-resolution.
In the scope of image processing expectation maximization (EM) algorithm takes conspicuous place among the other clustering techniques. EM algorithm is suitable for multidimensional data but it requires a number of clusters to be defined a priori that might be a problem for particular applications. The main aim of this paper is to provide time effective EM clustering modification in the case of the unknown number of clusters and multidimensional input. Our work is based on statistical histogram based expectation maximization algorithm (SHEM) proposed by Yang and Huang with the predefined number of clusters. This method utilizes the histogram to provide EM iterations. However, the estimation of the histogram becomes time consuming task with the increase of input data dimension. Our algorithm extends the use of SHEM algorithm by means of a hierarchical histogram data structure, which allows us to reduce the computational load in the multidimensional case as well as to provide an initialization in the case of the unknown number of clusters. The paper includes several experimental results demonstrating the advantages and the disadvantages of the proposed solution
In paper a method of atmospheric correction of hyperspectral images is proposed. On the first stage, observed image is used to obtain parameters of atmospheric distortions using common radiative transfer model. In contrast to other existing approaches we use full nonlinear form of the radiative transfer model and linear spectral model, which is applied to describe undistorted hyperspectral pixels. The combination of both models allows us to evaluate parameters of atmospheric distortions using only hyperspectral image and qualitative information about the scene. The latter is a list of spectral signatories (undistorted), which can appear in different linear combinations in the registered scene. The proposed method does not require any precedential information (sets of pixels containing predefined information) or pure hyperspectral pixels. Thus, it can be applied for blind identification of the atmospheric distortion model and for further atmospheric correction. Experimental results presented in this paper demonstrate performance of the method.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.