Vol.5 No.3 2012

9/94

Research paper : Developing an evaluation system of visually induced motion sickness for safe usage of moving images (H. Ujike)−144−Synthesiology - English edition Vol.5 No.3 (2012) Therefore, the biomedical effect occurred when the velocity of the visual global motion in the image was in the range where the biomedical effect level shown in the vertical axis in Fig. 5 reached a certain level.[11]The point to notice in the VIMS experiment is the adaptation effect to the VIMS, and to reduce the effect, it was necessary that the participant of the experiment took at least a two-week interval before participating in the next experimental session. The combination condition in Fig. 5 is when the adaptation effect was minimized as much as possible, but with over 40 participants and with the requirement of keeping two-week intervals between sessions, over half a year was necessary to obtain the results.4.2 Visual global motion analysis of the imageIf the types of visual global motion and the temporal variation in the image subject to evaluation are clarified, it will be possible to evaluate the degree of VIMS that may be induced, based on the findings in the previous section. Therefore, in the VIMS evaluation system, it is necessary to analyze the visual global motion of the subject image, extract the components and the speed of motion contained in the image, and to calculate the temporal variation of speed. The researches of the image compression technology can be applied to realize the above.[12] Therefore, we selected this technology as the compositional element of the VIMS evaluation system, and had people and companies that specialized in the image compression technology participate in the R&D phase.There are two phases in the process of the visual global motion analysis technology. In the first phase, the image region is divided into, for example, 16 rows × 16 columns, and the place to which each region transferred in the next frame is searched by pattern matching. The amount and direction of transfer are calculated as the motion vector (local motion vector = LMV) of each region. In the second phase, the LMV component of each region is subject to cluster analysis, and the global motion vector corresponding to the basic motion of the camera such as pan, tilt, roll, or zoom is calculated. This corresponds to the visual global motion in this system.The issue of trade-off between the analysis time and analysis accuracy must be considered for this visual global motion analysis technology. In the initial phase of the system construction, about 15 second analysis time per frame of image was necessary, since we increased the accuracy of the system as a whole by incorporating the method with highly accurate motion analysis to improve each technological element. However, considering practical use, it was desirable that the speed of analysis be about the same as the replay time. Therefore, in the final stages of the development, we improved the method of LMV calculation to reduce the analysis time, and also checked that there were no practical issues in the analysis accuracy, and realized the practical analysis technology for visual global motion. This analysis technology was realized by Hitachi Consulting Co., Ltd.4.3 VIMS evaluation modelIn the VIMS evaluation system, the core part was the input of the temporal variation of speed and the types of the visual global motion in the image, and the output was the temporal variation of the degree of VIMS that may occur.[13] Since the assumption based on the basic characteristic of VIMS was necessary between the input and output, we created the VIMS evaluation model to realize this. The VIMS evaluation model was important in the following two perspectives.(1)While this system assumes the temporal variation of the degree of VIMS, there is no temporal factor included in the findings of biomedical effects explained in subchapter 4.1. Therefore, it is necessary to consider the effect of the display time of images to construct a model including this factor.(2)While this system aims for the estimation of VIMS in general images including the complex visual global motion, the findings of biomedical effects explained in subchapter 4.1 use simple visual global motion only. It is necessary to investigate whether it is possible to estimate the degree of VIMS in general images using this basic finding.Therefore, from the perspective of (1), we investigated the variation of discomfort due to VIMS when the speed of visual global motion in the image is included in the speed range that may cause VIMS. As a result, it became clear that, due to the existence of the visual global motion that corresponded to that speed range,[13] there were transient responses in which discomfort increased in a short time, and sustained responses where the discomfort did not decrease for a while even after the disappearance of the visual global motion. Therefore, in the VIMS evaluation model, the transient and the sustained responses are output when each type of visual global motion are included in the corresponding speed range for a certain time.Next, from the perspective of (2), to increase the accuracy of the VIMS evaluation model, the image introduced in subchapter 4.4 as likely to induce VIMS was prepared to calculate the temporal variation of the degree of VIMS by using the VIMS evaluation model. Then, the biomedical effects were measured by having the participants view the image using the same image in the procedure explained in subchapter 4.5, and the two results were compared. The parameters of the VIMS evaluation model were adjusted based on the results.

※このページを正しく表示するにはFlashPlayer9以上が必要です