In Japan, a sustainable supply of rice should be guaranteed since it is the staple food. Therefore it is important to stabilize rice farmers’ incomes by giving them appropriate compensation even at the time of adverse weather, e.g. cold summer. In the case of compensating farmers’ incomes based on a comparison between the usual yield and estimated yield, an accurate and quick estimation of rice yield is significant to help farmers at the time of disaster. However, a manual survey requires a huge amount of effort and time to investigate all the fields where damage has been declared, especially when a largescale cool summer occurs; therefore, we propose a yield estimation system using satellite images and part of the yield data. Our system provides an accurate and quick yield estimation for the vast extent of the fields at a reasonable cost. Many rice yield estimation methodologies utilizing satellite images have been studied over the years. For example, a crop growth model-based method and deep learning based method that utilize many satellite images taken multiple times at different times have been proposed. Although we can get much information about fields from many satellite images, systems using plural images cost much and need exhaustive calibration before the estimation. Therefore, in this study we propose a yield estimation method using a single image that is taken just before the harvest time. First, we extracted the spectral values from the satellite image using field GIS data and then used a mixed model to perform rice yield estimation. Mixed model is expanded linear regression model and able to take the difference between rice varieties, such as Yumepirika and Nanatsuboshi, into accounts. In addition, we introduced two vegetation indexes, normalized difference vegetation index (NDVI) and green normalized difference vegetation index (GNDVI), into our model as feature values. Generally, NDVI and GNDVI have a positive correlation with the volume of the plant on the field and are used for yield estimation. Of course, we could use machine learning methods, for example random forest and support vector regression. However, we adopted a mixed model considering the explainability of the results and tha fact that the number of input feature values is small. The area of interest is Asahikawa-City, Hokkaido Province, Japan. We demonstrated our method on two datasets and evaluated the performance of our model based on mean absolute error (MAE) using 10-fold cross-validation. One dataset was damaged field data acquired in 2018 (2170 fields). The other was undamaged field data acquired in 2017 (1358 fields). We used RapidEye and SPOT-6 satellite images in 2017 and 2018, respectively. Our experimental results show that our model reduces the MAE of the estimated yield by over 2.5% percent compared to conventional regression methods in each damaged field and undamaged field case.
We describe in this paper how we mixed 3D information, i.e., a DSM (Digital Surface Model), for segmentation tasks in airport environments. The segmentation output classes were set to asphalt, concrete, and building classes because these are informative for distinguishing airport functionality. DSM is very informative for extracting buildings because airports are usually located on flat fields; however, high resolution DSMs are not provided for free. Therefore, we gathered adequate numbers of very-high-resolution satellite images and generated DSMs through stereo processing by ourselves. At the same time, we trained a modified U-NET for the initial segmentation. By leveraging the results of the segmentation, we identified ground pixels, i.e., asphalt or concrete, and calculated the ground height. Then, we applied an adaptive threshold algorithm to the DSMs by using the ground height and extracted building masks. Finally, we concatenated probability maps from the modified U-NET and building masks that represented the building class with a high precision in the flat airport fields. Consequently, we obtained better performance than the initial segmentation results, especially in the case of the building class. In experiments, we confirmed that the modified U-NET could detect asphalt and concrete with a high precision and that it was possible to identify ground pixels and extract building masks. The performance of our approach was improved by 20%, especially in detecting the building class. For future work, we will improve the quality of stereo processing and combine size specific detectors to achieve more accurate detection.
Position information of unmanned aerial vehicles (UAVs) and objects is important for inspections conducted with UAVs. The accuracy with which changes in object to be inspected are detected depends on the accuracy of the past object data being compared; therefore, accurate position recording is important. A global positioning system (GPS) is commonly used as a tool for estimating position, but its accuracy is sometimes insufficient. Therefore, other methods have been proposed, such as visual simultaneous localization and mapping (visual SLAM), which uses monocular camera data to reconstruct a 3D model of a scene and simultaneously estimates the trajectories of the camera using only photos or videos. <p> </p>
In visual SLAM, UAV position is estimated on the basis of stereo vision (localization), and 3D points are mapped on the basis of the estimated UAV position (mapping). Processing is implemented sequentially between localization and mapping. Finally, all the UAV positions are estimated and an integrated 3D map is created. For any given iteration in the sequential processing, there will be estimation error, but in the next iteration, the previous estimated position will be used as a base position regardless of this error. As a result, error accumulates until the UAV returns to a location it passed before. Our research aims to mitigate this problem. We propose two new methods. <p> </p>
(1) Accumulated error caused by local matching with sequential low-altitude images (i.e. close-up photos) is corrected with global-matching between low- and high-altitude images. To perform global-matching that is robust against error, we implemented a method wherein the expected matching areas are narrowed down on the basis of UAV position and barometric altimeter measurements. <p> </p>
(2) Under the assumption that absolute coordinates include axis-rotation error, we proposed an error-reduction method that minimizes the difference in the UAVs’ altitude between the visual SLAM and sensor (bolometer and thermometer) results. <p> </p>
The proposed methods reduced accumulated error by using high-altitude images and sensors. Our methods improve the accuracy of UAV- and object-position estimation.
Unmanned aerial vehicles (UAVs) are being used to reduce the cost and risk of facility inspections. For the power distribution companies, power line inspection for providing stable power supply is an important but costly task. It includes deterioration diagnosis, detection of foreign matter adhesion, and estimation of power line-tree conflict risk, all of which is currently performed visually on foot. In this study, we explore the methods of detection and visualization of a power line-tree conflict using aerial images taken by drones. To detect a power line-tree conflict, we should firstly recognize the power lines and trees in the aerial images in order to identify the “candidate” regions of the conflict, and secondly, estimate the actual positional relationship between them in 3D. However, as previous studies have shown, the detection of power lines in an image is a challenging task because they are very narrow and monochromatic, which results in difficulty in extracting features. This specific character of the power lines could also cause failure in 3D reconstruction, in which feature matching among images is necessary. Here, we show that convolutional neural networks (CNNs) can be effectively applied in recognition of power lines and trees in an image. We also found that in mapping the candidate region of conflict to a 3D model the power line position could be estimated by taking the pole height into account. This way, even if it is difficult to reconstruct the power line in 3D, a user can make the final decision about the conflict by checking the depth and/or the height directional relationship.
Bathymetry at shallow water especially shallower than 15m is an important area for environmental monitoring and
national defense. Because the depth of shallow water is changeable by the sediment deposition and the ocean waves, the
periodic monitoring at shoe area is needed. Utilization of satellite images are well matched for widely and repeatedly
monitoring at sea area. Sea bottom terrain model using by remote sensing data have been developed and these methods
based on the radiative transfer model of the sun irradiance which is affected by the atmosphere, water, and sea bottom.
We adopted that general method of the sea depth extraction to the satellite imagery, WorldView-2; which has very fine
spatial resolution (50cm/pix) and eight bands at visible to near-infrared wavelengths. From high-spatial resolution
satellite images, there is possibility to know the coral reefs and the rock area’s detail terrain model which offers
important information for the amphibious landing. In addition, the WorldView-2 satellite sensor has the band at near the
ultraviolet wavelength that is transmitted through the water. On the other hand, the previous study showed that the
estimation error by the satellite imagery was related to the sea bottom materials such as sand, coral reef, sea alga, and
rocks. Therefore, in this study, we focused on sea bottom materials, and tried to improve the depth estimation accuracy.
First, we classified the sea bottom materials by the SVM method, which used the depth data acquired by multi-beam
sonar as supervised data. Then correction values in the depth estimation equation were calculated applying the
classification results. As a result, the classification accuracy of sea bottom materials was 93%, and the depth estimation
error using the correction by the classification result was within 1.2m.
Interpretation of high-resolution satellite images has been so difficult that skilled interpreters must have checked the satellite images manually because of the following issues. One is the requirement of the high detection accuracy rate. The other is the variety of the target, taking ships for example, there are many kinds of ships, such as boat, cruise ship, cargo ship, aircraft carrier, and so on. Furthermore, there are similar appearance objects throughout the image; therefore, it is often difficult even for the skilled interpreters to distinguish what object the pixels really compose. In this paper, we explore the feasibility of object extraction leveraging deep learning with high-resolution satellite images, especially focusing on ship detection. We calculated the detection accuracy using the WorldView-2 images. First, we collected the training images labelled as “ship” and “not ship”. After preparing the training data, we defined the deep neural network model to judge whether ships are existing or not, and trained them with about 50,000 training images for each label. Subsequently, we scanned the evaluation image with different resolution windows and extracted the “ship” images. Experimental result shows the effectiveness of the deep learning based object detection.