Convective storms are dangerous atmosphere hazards often accompanied by heavy rain and strong wind. Despite their
short life time and small spatial scale, Doppler weather radars can provide a 3-D high temporal and spatial resolution of
continuous measurement. Thus, a combination of identification, tracking and nowcasting of convective storms based on
radar data is one of the most important meteorological methodologies. However, most forecasting algorithms only use
extrapolating techniques and do not reflect storm shape changes, making forecasting result unreliable. Thus, this paper
presents a new method, namely level set, to forecast storms with storm cell shape changes concerned. As a result,
forecasting precision is improved.
A two-dimensional grid plate can offer an X-Y position standard where grids are aligned orthogonal to each other. It is
important to ensure the positional accuracy of the grid plate when the grid plate is used to calibrate planar movement
systems, such as vision measuring machines and scanning probe microscopes. Existing algorithms for self-calibration
employ the discrete Fourier transform, which is complicated and has poor noise suppression capability. We have
developed an algorithm that can achieve exact self-calibration for a two-dimensional grid plate using the least squares
method when there is no random noise. In the presence of random noise, the algorithm still presents an excellent
capability for noise suppression. As an extension of the classic three-location measurement, the algorithm can be applied
to four- or five-location measurements, which reduce measurement uncertainties. The error propagation characteristic of
the random errors has been investigated in the case of different measurement strategies. According to the simulation
results, the mean error propagation ratios are less than 1 when the array size of the grid plate is less than 32×32. Finally,
the influence of the scale errors of the planar movement system is discussed.
The template updating problem of Kernel-based tracking (KBT) includes two aspects: target-scale update and target-model update. The proposed algorithm can update both tracking window's scale and target model by making use of continuous adaptive distribution. The ability of KBT can be complemented within its own framework with modest computation cost. The proposed tracking algorithm tries to get a balance between the stability of KBT and adaptability of CAMSHIFT for creating a robust tracker.
In order to build a content-based cloud image retrieval system, a deformable circle model is proposed to represent cloud
shape. After a description of the model and shape decomposition process, a hierarchical similarity rule is presented. Then
comparative experiments with some conventional contented-based image retrieval methods are carried out, which leads
to a finding that the retrieval performance of our proposed method is superior to others.