Shengyang Li

An automatic MTF measurement method for remote sensing cameras

The modulation transfer function (MTF) for remote sensing imaging sensors sometimes need be automatically measured both in order to fast calibrate them for best performance while working on board and to do (near) real-time digital compensation for inevitable degradation. In this paper, we presented such an automatic MTF measurement approach. It is based on the detection of straight lines with Hough transforms. Once qualified lines are determined, the MTF is estimated by a step edge method. Experiments with IKONOS images showed the proposed technique has a high success rate for automatically measuring the MTF with an acceptable accuracy.

Image inpainting based on anisotropic MRF model

The goal of image inpainting is to restore the damaged or missing pixels on images and it is an active research topic in image engineering. In order to restore narrow gaps on damaged images, we propose a type of anisotropic inpainting model based on Markov Random Fields. The inpainting model can preserve the edges and orientational texture. We implement our method using Simulated Annealing algorithm. Experiments show that the proposed method can obtain satisfying results and is practical in applications.

Low-Quality and Multi-Target Detection in RSIs

To improve the recognition and detection accuracy and recall of objects in unnatural images, we use remote sensing video image data to detect low-quality and small targets. Based on this, we propose the use of a deconvolution network and hyper features to control convolutional feature quality. We call this approach the Quality Deconvolution Single Shot Detector (QDSSD) detection model. Through the frame-by-frame annotation of the video data from Jilin Satellite No. 1, we propose a CSU-RSI-Video dataset, with no fewer than 30 targets per image frame. We published the data for researchers to do experiments. To achieve small target detection, we enrich the information by gradually adding the underlying detail features to the upper layers and deconvolve the high-layer information to obtain stable detailed features for target detection. The empirical results show that, among the detected problems of low-quality small targets, the improved QDSSD network has better detection capability, and the detection effect is the best for many small targets that are close to each other. In the CSU-RSI-Video dataset, the QDSSD model’s mAP achieves 0.90227 for a single target. Comparing to the You Only Look Once (YOLO) model, the result is still superior in accuracy.

Mapping High Mountain Lakes Using Space-Borne Near-Nadir SAR Observations

Near-nadir interferometric imaging SAR (Synthetic Aperture Radar) techniques are promising in measuring global water extent and surface height at fine spatial and temporal resolutions. The concept of near-nadir interferometric measurements was implemented in the experimental Interferometric Imaging Radar Altimeters (InIRA) mounted on Chinese Tian Gong 2 (TG-2) space laboratory. This study is focused on mapping the extent of high mountain lakes in the remote Qinghai–Tibet Plateau (QTP) areas using the InIRA observations. Theoretical simulations were first conducted to understand the scattering mechanisms under near-nadir observation geometry. It was found that water and surrounding land pixels are generally distinguishable depending on the degree of their difference in dielectric properties and surface roughness. The observed radar backscatter is also greatly influenced by incidence angles. A dynamic threshold method was then developed to detect water pixels based on the theoretical analysis and ancillary data. As assessed by the LandSat results, the overall classification accuracy is higher than 90%, though the classifications are affected by low backscatter possibly from very smooth water surface. The algorithms developed from this study can be extended to all InIRA land measurements and provide support for the similar space missions in the future.

Object-oriented crops classification for remote sensing images based on convolutional neural network

Deep learning technology such as convolutional neural networks (CNN) has achieved outstanding results in the field of crops classification for remote sensing images. The way of land cover or crop types remote sensing classification using CNN is mainly pixel-based classification which is often affected by the phenomenon of “salt and pepper”. In order to reduce this effect, an object-oriented crops classification method based on CNN is proposed in this paper. By combining image segmentation technology and CNN model, we use this method to obtain the results of crops classification from Sentinel-2A multi-spectral remote sensing images in Yuanyang County, Henan Province, China. The experiment show that, compared with the pixel level classification based on CNN which only consider the spectral and temporal characteristics of the crops, the method we proposed comprehensively utilizes more detailed information such as spectral feature, texture feature, spatial relationship, and color space. Thus, it gains a better discriminability for some specific crop and achieves higher classification accuracy.

Key Technology Research of Massive Multi-source Heterogeneous Spatial Data Visualization and Management System Based on 3D Digital Earth

Its a key problem how to efficiently organize and manage the growing massive multi-source heterogeneous spatial science data. The paper researched a number of key technologies of data organization and management systems including data storage, data retrieval, and data visualization. Firstly, it proposed and designed effective system architecture, secondly, it enabled efficient storage, search and location of massive heterogeneous data by spatial relational database and distributed database technology, lastly, it improved the visualization efficiency of space science data by three-dimensional data earth, the levels of detail (LOD), the three dimensional data cutting, and multi-threaded parallel loading techniques. Based on the current spatial data management technology and structural features of space science data, we completed the integration of the massive multi-source heterogeneous data visualized organization and management system. The system in the application confirms the validity of these technologies for data storage, data retrieval and data visualization.

Multi-source high resolution remote sensing image fusion based on intelligent decision

Application fields of multisource high resolution remote sensing image fusion are expending ceaselessly. Especially, commercial remote sensing satellite data and image fusion algorithms are constantly emerging. This paper proposed an intelligent decision strategy that can be used to query various remote sensing data for recommending the optimal image fusion data combination and appropriate fusion algorithm. The system architecture was defined as four layers including interaction layer, service layer, component layer, and data layer. In order to extend more plug-ins, we introduced the Zero-C distributed service framework to invoke different types of service interface. Besides, a main interface of intelligent inference system is developed for users to submit image fusion tasks.

High resolution remote sensing image fusion method based on curvelet and HCS

To obtain high spatial and spectral resolution image, we propose a novel method of high resolution remote sensing image fusion based on the second generation curvelet transform and hyperspherical color space transform which can fuse n-band multispectral and panchromatic images. The GF-1 satellite images are used as experimental data, and the fused image are quantitatively analyzed according to the mean, the standard deviation, the correlation coefficient, the information entropy and the average gradient. The results show that the proposed method has better performance than other fusion methods such as Principal Component Analysis, Gramm-Schmidt, and Hyperspherical Color Sharpening.

Reliable clarity automatic-evaluation method for optical remote sensing images

Image clarity, which reflects the sharpness degree at the edge of objects in images, is an important quality evaluate index for optical remote sensing images. Scholars at home and abroad have done a lot of work on estimation of image clarity. At present, common clarity-estimation methods for digital images mainly include frequency-domain function methods, statistical parametric methods, gradient function methods and edge acutance methods. Frequency-domain function method is an accurate clarity-measure approach. However, its calculation process is complicate and cannot be carried out automatically. Statistical parametric methods and gradient function methods are both sensitive to clarity of images, while their results are easy to be affected by the complex degree of images. Edge acutance method is an effective approach for clarity estimate, while it needs picking out the edges manually. Due to the limits in accuracy, consistent or automation, these existing methods are not applicable to quality evaluation of optical remote sensing images. In this article, a new clarity-evaluation method, which is based on the principle of edge acutance algorithm, is proposed. In the new method, edge detection algorithm and gradient search algorithm are adopted to automatically search the object edges in images. Moreover, The calculation algorithm for edge sharpness has been improved. The new method has been tested with several groups of optical remote sensing images. Compared with the existing automatic evaluation methods, the new method perform better both in accuracy and consistency. Thus, the new method is an effective clarity evaluation method for optical remote sensing images.