arXiv subjects

Ryosuke Nakamura

Publications and source records attributed to Ryosuke Nakamura.

21 records · Page 2Linked to original sources

arXiv2017

Object Detection of Satellite Images Using Multi-Channel Higher-order Local Autocorrelation

The Earth observation satellites have been monitoring the earth's surface for a long time, and the images taken by the satellites contain large amounts of valuable data. However, it is extremely hard work to manually analyze such huge data. Thus, a method of automatic object detection is needed for satellite images to facilitate efficient data analyses. This paper describes a new image feature extended from higher-order local autocorrelation to the object detection of multispectral satellite images. The feature has been extended to extract spectral inter-relationships in addition to spatial relationships to fully exploit multispectral information. The results of experiments with object detection tasks conducted to evaluate the effectiveness of the proposed feature extension indicate that the feature realized a higher performance compared to existing methods.

cs.CV↗

arXiv2017

Solar Power Plant Detection on Multi-Spectral Satellite Imagery using Weakly-Supervised CNN with Feedback Features and m-PCNN Fusion

Most of the traditional convolutional neural networks (CNNs) implements bottom-up approach (feed-forward) for image classifications. However, many scientific studies demonstrate that visual perception in primates rely on both bottom-up and top-down connections. Therefore, in this work, we propose a CNN network with feedback structure for Solar power plant detection on middle-resolution satellite images. To express the strength of the top-down connections, we introduce feedback CNN network (FB-Net) to a baseline CNN model used for solar power plant classification on multi-spectral satellite data. Moreover, we introduce a method to improve class activation mapping (CAM) to our FB-Net, which takes advantage of multi-channel pulse coupled neural network (m-PCNN) for weakly-supervised localization of the solar power plants from the features of proposed FB-Net. For the proposed FB-Net CAM with m-PCNN, experimental results demonstrated promising results on both solar-power plant image classification and detection task.

cs.CV↗

arXiv2009

The Hayabusa Spacecraft Asteroid Multi-Band Imaging Camera: AMICA

The Hayabusa Spacecraft Asteroid Multiband Imaging Camera (AMICA) has acquired more than 1400 multispectral and high-resolution images of its target asteroid, 25143 Itokawa, since late August 2005. In this paper, we summarize the design and performance of AMICA. In addition, we describe the calibration methods, assumptions, and models, based on measurements. Major calibration steps include corrections for linearity and modeling and subtraction of bias, dark current, read-out smear, and pixel-to-pixel responsivity variations. AMICA v-band data were calibrated to radiance using in-flight stellar observations. The other band data were calibrated to reflectance by comparing them to ground-based observations to avoid the uncertainty of the solar irradiation in those bands. We found that the AMICA signal was linear with respect to the input signal to an accuracy of << 1% when the signal level was < 3800 DN. We verified that the absolute radiance calibration of the AMICA v-band (0.55 micron) was accurate to 4% or less, the accuracy of the disk-integrated spectra with respect to the AMICA v-band was about 1%, and the pixel-to-pixel responsivity (flatfield) variation was 3% or less. The uncertainty in background zero-level was 5 DN. From wide-band observations of star clusters, we found that the AMICA optics have an effective focal length of 120.80 \pm 0.03 mm, yielding a field-of-view (FOV) of 5.83 deg x 5.69 deg. The resulting geometric distortion model was accurate to within a third of a pixel. We demonstrated an image-restoration technique using the point-spread functions of stars, and confirmed that the technique functions well in all loss-less images. An artifact not corrected by this calibration is scattered light associated with bright disks in the FOV.

astro-ph.EP↗