Hajime Hara

Analysis and evaluation of tennis plays by computer vision

In this paper, tennis plays are analyzed, and skills of players are evaluated quantitatively by computer vision technology. Two cameras are used to detect the three-dimensional positions of balls and players. Subtraction of images are used to extract ball images eliminating ball shadows. The prediction of ball position by using velocity vectors of previous images enables accurate and high-speed ball detection. Sequential images of 64 return strokes and 15 services by four players were analyzed quantitatively. Six parameters that represent tennis play skill are introduced for the quantitative evaluation. These skill factors are evaluated using the multiple regression method. Among these, it is concluded that ball scattering and ball shot speed are strongly related to skill scores, and the skill factors varies between players.

Analysis of signal propagation in 1-D CNNs with the antisymmetric template

This paper studies some fundamental properties on signal propagation in one-dimensional cellular neural networks with the antisymmetric template under the assumption that the initial output has only one connected component. A new sufficient condition for the component to propagate without attenuation is derived through theoretical analysis. It is also proved that under the same condition the final output is a black-and-white image having only one black at the right-most pixel, which means that the network can perform connected component detection when the input image has only one connected component.

Necessary and Sufficient Conditions for One-Dimensional Discrete-Time Autonomous Binary Cellular Neural Networks to Be Stable

We give the necessary and sufficient conditions for a one-dimensional discrete-time autonomous binary cellular neural networks to be stable in the case of fixed boundary. The results are complete generalization of our previous one [16] in which the symmetrical connections were assumed. The conditions are compared with some stability conditions so far known.

Necessary and Sufficient Conditions for a 1-D DBCNN with an Input to Be Stable in terms of Connection Coefficients

We give necessary and sufficient conditions for a 1-D DBCNN (1-dimensional discrete-time binary cellular neural network) with an external input to be stable in terms of connection coefficients. The results are generalization of our previous one [18], [19] in which the input was assumed to be zero.

Necessary and sufficient conditions for one-dimensional discrete-time binary cellular neural network with nonsymmetric connections to be stable

We give the necessary and sufficient conditions for a one-dimensional discrete-time binary cellular neural networks with non-symmetric connection to be stable. The results are complete generalization of our previous one, in which the symmetrical connections were assumed. The conditions are compared with the stability condition recently derived by the authors.

Pattern separation function of two-layered random net with feedforward inhibitory connections

The two-layered random neural network with the feedforward inhibitory connections has the interesting property that the input patterns with the same firing rate are separated uniformly under a certain condition, independently of the overlapping rate of the patterns. This paper gives the general condition for the uniform separation of the input patterns with unequal firing rates, independently of the overlapping rate. Then, a detailed condition is derived from the general condition. Thus it is shown that the network has the function under a certain condition to separate uniformly the input patterns with unequal firing rates. In parallel to this, the effects of the firing rate of the second layer, the firing rates of the input patterns and their ratio on the pattern separation function, are investigated. Based on the investigation, it is shown that the separating function is affected greatly by those firing rates and their ratio.