Vol.5 No.3 2012

10/94

Research paper : Developing an evaluation system of visually induced motion sickness for safe usage of moving images (H. Ujike)−145−Synthesiology - English edition Vol.5 No.3 (2012) 4.4 Production of images likely to cause sickness based on the image production methodFor the images that contain the potential factors of VIMS to be used in improving the accuracy of the VIMS evaluation model, the production was subcontracted to the image producers who are experts of image production. For the production of such images, it was possible to use the CG images of simple contents containing complex visual motion or live-action images that were shot by the researchers, but these might not be necessarily acceptable to the experts of image production. For example, a non-expert image that contains potential factors of VIMS may not necessarily contain the visual effects and methods that are utilized by the experts, and may not be convincing to the image producers. This is not necessarily easy to explain in terms of logic, and is beyond the bounds of technological discussion, but it may be an important point in gaining understanding and cooperation on the image safety from the image industry people.As conditions for the image production, in addition to the inclusion of potential factors of VIMS, the condition was the use of live-action images that could be assumed as the actual subject of evaluation. In the former perspective, to investigate the effect of the motion speed of pan, tilt, roll, and zoom that are basic motions of the camera, we attempted to include those potential factors based on the findings pertaining to the basic characteristic of VIMS. However, they had to be live-action images from the latter perspective, and the speed of the motion of the camera was difficult to measure at the shooting session, and the following procedure was used upon discussion with the image producer. First, five-step speed was set for each of the basic camera motion. Specifically, for pan, tilt, and roll, the settings were: very slow 7.5 deg/s, slow 15 deg/s, medium 30 deg/s, fast 60 deg/s, and very fast 80 deg/s. In case of zoom, the magnifications (and reductions) between the frames were: very slow 1.15 (0.86), slow (1.30 0.77), medium 1.50 (0.67), high 1.75 (0.57), and very fast 2.00 (0.50). Next, these speeds were held constant by special effects (SFX) shots, and motion was in one direction for the first 8 seconds, in the opposite direction for the next 8 seconds, and the basic image of 16 seconds in both directions was created. Since there were five-step speeds for the four types of basic movements, there were a total of 20 types of images. The motion of the camera used in shooting each scene was matched with the motion of the camera of the respective basic images, so the final live imagery contained the scenes that corresponded to each speed expressed in 20 types of basic images.The important point in these images was that the participants of the experiment could view the images without becoming bored during the experiment, and so a minimum plot was added to the image. However, it was necessary to keep the plot to a minimum to prevent the story from providing emotional effects that may affect the biomedical measurements. Therefore, the cooperation of the image producer was essential.4.5 Biomedical effect measurement to increase the accuracy of the VIMS evaluation modelIn the biomedical effect measurement for VIMS, in general, there are psychological measurements that rely on subjective evaluation and the physiological measurements that investigate the effect on the autonomous nervous system. In the previous researches, the emphasis was mainly on either one of the measurements, and there was hardly any investigation on the relationship of the two measurements for temporal variation of VIMS. However, subjective evaluation requires supportive evidence of objective data as much as possible, since there is an issue of propensity in the individual differences. On the other hand, in physiological measurement, it is necessary to match with the subjective evaluation to investigate what the variations in the measurements indicate. Therefore, the two measurements were mutually essential to increase the reliability of the measurement data of VIMS.In the development of the VIMS evaluation system, the psychological and physiological measurements were conducted simultaneously, and the investigations were done for the relationship of the temporal variation of the two. For the physiological measurement, we obtained cooperation of the Tohoku University, Niigata University, and Fukushima University that had plentiful experiences in developing the measurement method, and we investigated the relationship of the temporal variation of the subjective evaluation and the physiological indices during the image viewing.[14]-[17] To execute this experiment, common experimental protocols were used and the measurement data were shared among the above universities and AIST, and this enabled efficient collection of data from several experimental participants. Through such biomedical effect measurement, the Tohoku University and Fukushima University showed that the max, which is the maximum cross correlation function between the blood pressure and heart rate, changed with about 1 minute gap along the temporal variation of subjective evaluation, and that it was possible to estimate the temporal variation of subjective evaluation using the multiple physiological indices obtained by the measurements of electrocardiogram and photoplethysmogram.[14][17]In the VIMS evaluation model, as discussed in subchapter 4.3, the specification was that it would output the transient and sustained responses each time the visual global motion was included in the corresponding speed range for a certain time. For the output of the model, the accuracy was improved by approximating it to the impulse response function estimated from the time-series data of the biomedical effect measurement conducted using several images explained in

※このページを正しく表示するにはFlashPlayer9以上が必要です