Image Processing (2011-)


Automatic Grading System for Programming Contest (2008-)

This is a project to produce a system to grade submitted programs with measuring computation time and memory consumption. The goal is not only to protect the system from illegal programs but also to grade various tasks such as interactive tasks.

Related Paper

  • Kentaro Imajo, "Contest Environment Using Wireless Networks: A Case Study From Japan," Olympiad in Informatics 5, 26―31, 2011.

Short-term Forecast of Precipitation Using an Optical Flow Algorithm (2011)

We implemented a program to forecast precipitation in the future with the present transition generated by matching previous images. The proposed method gets 25 percent higher accuracy than the forecast of precipitation Nowcast, which is announced by Japanese Meteorological Agency. Additionally, the computational time is short enough to forecast precipitation with a more detailed map minute by minute.
Figure 1: Radar map
at 2011-01-30 00:00
Figure 2: Radar map
at 2011-01-30 00:55
Figure 3: Forecast of Nowcast
for 2011-01-30 00:55
Figure 4: Forecast of the proposed method for 2011-01-30 00:55
(Gray area indicates that the area was originally outside at 00:00.)
Figure 4 is the result of the proposed method, and it is the forecast of 55 minutes later based on Figure 1. We can see that Figure 4 has higher accuracy than the forecast of Nowcast Figure 3 that was announced at 2011-01-30 00:00. It seems that the difference is caused by lack of local motion prediction of Nowcast. Basically, Nowcast could not forecast that a rainy area did not go beyond mountains or the difference of the motion between the Sea of Japan and the Pacific Ocean was bigger than Nowcast had expected.

Incidentally, the previous methods to detect optical flows do not have enough precision to forecast, so I developed and implemented a new algorithm. The algorithm can treat images that are taken with a stereo camera as if they know 3D information using interpolation and extrapolation with the optical flows.
Figure 5: Left-side view
Figure 6: Right-side view
Figure 7: Extrapolation from Figure 6 with three times distance
between Figure 5 and Figure 6
Figure 5 is the left-side view picture of a stereo camera, and Figure 6 is the right-side view picture of a stereo camera. Figure 7 can be produced with the images by extrapolating the optical flow between Figure 5 and Figure 6 three times. Moreover, we can also produce movies such as [GIF animation generated by an optical flow algorithm] with interpolation and extrapolation of only two images Figure 5 and Figure 6.

Related Paper

  • Kentaro Imajo, Go Hasegawa, Yoshiaki Taniguchi, Hirotaka Nakano, "Short-term Precipitation Forecasting Using an Optical Flow Algorithm," IEICE Technical Report 110(455), 93-98, 2011.

Continuous Wavelet Transform (2009-)

Related Paper

  • Kentaro Imajo, "Computation for the Fast Continuous Wavelet Transform," JSIAM, 375-376, 2009.