A Representation of Human Visual Acuity on an Immersive Projection Display Yukihiro Ishibashi and Mitsunori Makino Graduate School of Science and Engineering, Chuo University, Japan E-mail: yishibas@makino.ise.chuo-u.ac.jp, makino@m.ieice.org Abstract This paper proposes a representation method of a scene influenced by weak or different human vision on an immersive projection display (IPD) such as the CAVE. Recently reduced visual acuity has been social problem, especially for children, in Japan. Myopia, short sight, is major case among causes for the reduced visual acuity. Since both of early detection and treatment are important for visual restoration, educational activities are necessary. Among the activities, hands-on experience is effective especially for children, since the experience shows how unclearly they can see a scene in the case of weak and/or different vision. Furthermore, it is much effective that they can see any scene from any viewpoint. Therefore in this paper, we propose a visualization method of a scene on IPD, in which human visual acuity is considered. Displaying the images on IPD supported by the CAVElib, the method shows the user stereoscopic blurred target objects far from focal points of the eyes, while it shows clear objects near the points. As a result, the proposed method is useful for children as an educational hands-on tool of human vision. 1 Introduction Recently reduced visual acuity has been social problem, especially for children, in Japan. Myopia, short sight, is a major case among causes for the reduced visual acuity. Although both of early detection and treatment are important for visual restoration, symptom of myopia is easily overlooked. Therefore, educational activities are necessary for children. Among the activities, hands-on experience is effective especially for children. Through experiencing scene which they have not experienced before, children are encouraged understanding and consciousness[1]. However, they cannot experience three dimensional sense, as well as they cannot interactively change their viewpoints and experience the influence with several visual acuities on textbooks and/or video. Also textbooks and/or video are not rendered case of different visual acuity on each eye. CAVE is known as a typical immersive projection display (IPG) system. CAVE needs two images for each eye to make a three dimensional image. Since CAVE can render different visual acuity in three dimension, CAVE is expected to be useful to educational tool for myopia. In this paper, we aim to educational support for show children a scene influenced scene by visual acuity for children. Therefore we propose a visualization method of a scene on IPD, in which human visual acuity is considered. 2 Visual Feature 2.1 Structure of Human s eye and Myopia According to Ref.[2], human s augen is a sphere with 24-25mm diameter. Cornea, iris and crystal lens are located anterior to the augen, while retina is located posterior to it (see Fig.1). Figure 1: Human eye cross-sectional view[3] Under health vision, the cornea and the lens focus incident light on the retina (see Fig.2). However, they happen to focus the light in front of the retina, when a far obfect is unclearly seen (see Fig.3). Such situation is defined as myopia, which is caused by a certain abnormal situation on the cornea, the lens and/or size of augen. Wearing glasses is effective as
its symptomatic treatment, as well as surgery of the retina or the lens is direct remediation. However, both of preservation and early detection of the myopia are the first priority for keeping healthy vision. Figure 4: Landolt ring(for 5m, visual acuity 1.0) Figure 2: Emmetropia condition[4] Figure 3: Myopia condition[4] 2.2 Landolt Ring for Static Vision Visual acuity is called as static vision, when both of eyes and a target object stand still. The Landolt ring, C-shaped black ring shown in Fig.4, is wellknown figure by which we can measure the static vision. By using different size of the rings, we determine the smallest size of the ring on which we can detect white missing area. From relationship between the size and distance from viewpoint, the smallest field of view under which we can distinguish two points. For example, suppose that the smallest ring is 7.5mm of diameter, 1.5mm of wire size, and 1.5mm of missing width, and that the white missing area can be detected 5m awayfrom the ring. Then, the field of view is 1 second (one-sixty degree). Therefore, the visual acuity becomes 1.0 since the acuity is defined as inverse number of the field of view (1 second)[5]. Under healthy vision the visual acuity becomes 1.0-1.2[6]. However, the visual acuity usually under the myopia decreases less than one under the healthy vision, since the incorrect focus causes hard detection of missing area. 3 Proposal Method 3.1 Requirement and Overview In order to construct a representation of myopic situation as an educational tool, the representation has to give us views according to light simulation from an object to retina through cornea and crystal lens. Also the views should be changed when the given visual acuity or allocation of the object is different. For example, the representation generates an unclear image of a far object for the given visual acuity, while the representation generates a clear image of a near object. Furthermore, the representation should handle difference of visual acuity for both eyes, since such situation often happens on myopic person. Therefore the representation requires a stereoscopic display system such as the CAVE, which gives us two views from both eyes simultaneously. In this section, we construct a representation of myopia on the CAVE with the CAVELib, a graphic library for the CAVE. At first, we define visual acuity for each eye. Then, we determine refraction index for each eye by calculating a focal point inside an augen from the given visual acuity. By using the index, we represent spread of blur (unclear view) for an object according to distance between the object and retina. 3.2 Calculation of Refraction Index Here let diameter of an augen be 24mm. Also let semimajor axis and semiminor axis of crystal lens be 5mm and 2mm, respectively (see Fig.5). According to the definition of the visual acuity 1.0 described in Sect.2.2, we define d max [m] as the longest length between an object and retina, under which human having visual acuity s can discriminate 1.5mm white area on the Landolt ring (see Fig.6). d max is calculated as the following formula: d max = 1.5 10 3 tan 1 s (1)
Figure 7: Light refraction and shorter focal length under myopia Figure 5: Augen Model Figure 6: Relationship between visual acuity s and length d max Suppose that an object is located d[m] away from retina. Under myopia, unclear view is given if d d max since focal length f is shorter than distance f 0 between the crystal lens and the retina. Othermise clear view is given so that f is equal to f 0. Let ε[m] be d d max. Considering the myopic situation for the augen model defined above, we define f[m] as follows: n = sin ϕ sin θ 3.3 Determination of Affected Area (4) For easy explanation, suppose that x-y plane where center of an eye is set to its origin point and representative point (center) of an object is set to P c (p x, 0) (see Fig.8). Then, refraction index n is calculated with the given visual acuity and P c by Eq.(4) in 3.2. By using n, we determine position Q(q x, q y ) where a light ray from P c intersects retina after refraction at the upper edge of the crystal lens (see the first path in Fig.8). Moreover, we determine position P b (p x, p y ) from where a light ray intersects Q on retina after refraction at the lower edge of the lens (see the second path in Fig.8). It is noted that a human with the given visual acuity cannot discriminate P c from p y -neighbor points such as P b, since light rays from the area overlap the same are on his/her retina. It causes a blurred image of the object represented by P c. In 3D space, a circle on y-z plane, which has center of P b and radius of p y, is regarded as affected area by myopia. f = (12.0 + 119.0)e ε (2) Here suppose that r is radius of a virtual sphere, which shares a part of surface with the crystal lens in the augen (see Fig.5). Then refraction angle ϕ at edge of the crystal lens is calculated by the following equation (see Fig.7): ϕ = tan 1 (r/5.0) tan 1 (f/5.0) (3) Figure 8: Affected area(s) Considering ϕ derived from Eq.(3) and incident angle θ, we determine refraction index n as follows:
3.4 Representation of Myopic View Let I o be a CG image generated by normal method for the given viewpoint and objects without focus. Suppose that each pixel on I o has color data, information on visible object, and distance between the viewpoint and the visible object. We note that I o represents an image on human retina. Also let I b be a CG image generated by the proposed method. I b has only color data and its color on each pixel is black initially. Taking the given visual acuity, we calculate affected length d max by Eq.(1). d max is compared with distance information for any pixel (k, l) on I o. If I o (k, l) represents no object or an object locating nearer than d max, pixel (k, l) is determined not to be affected by myopia. In this case, add color information of I o (k, l) to I b (k, l). Otherwise if I o (k, l) represents an object locating further than d max, pixel (k, l) is determined to be affected by myopia. In the affected case, affected pixel set {(i, j) k α i k + α, l α j l + α} is calculated based on discussion in 3.3. Then, color information of I b (i, j) is updated by the followings: representing objects being further than d max in 3.4. The computation causes difficulty of real-time and precise visualization of myopia situation according to the head-tracking. 4 Simulation and Evaluation 4.1 Simulation We implemented the proposed method on Chuo- CAVE, three-faces IPD with head tracking, and CRTbased IPD with no tracking, which displays stereoscopic view on a CRT under the same principle as the CAVE. For experiment, alphabetical characters are introduced to objects in a virtual scene (see Fig.9) since they are well seen in our real life and used for an eye chart[7], while the Landolt ring can be seen only under vision test. I b [i][j] = I b [i][j] + W (k, l, i, j)i o [k][l] (5) where, W (k, l, i, j) = sum = i,j distance (1.0 radius )2 sum (6) (1.0 distance radius )2 (7) Here distance is defined as distance from (i, j) to (k, l), and radius is radius of the affected area. Applying the above procedures to images from both eyes, we obtain two blurred images affected by myopia based on different visual acuities. Overlapping these images on IPD, we can see virtual world under myopia. 3.5 Features The proposed method can represent stereoscopic CG images affected by myopia according to the given viewpoint and visual acuity as well as the given scene. Therefore the method can help us easily understand influence by myopia. Moreover, tracking head positions periodically with support of the CAVElib, the proposed method can represent the situation under myopia more effectively. However, computation is almost directly proportional to resolution of generated image (i.e., number of pixels) and number of pixels Figure 9: An example of alphabetical characters in a virtual scene We set 9 letters of MAKINOLAB in the scene. Figures 10-14 show results under different situation. Figure 10 shows a clear image without focus, which is generated by normal CG rendering method. On the image we can see clearly all characters. Figure 11 shows an image from the left eye with 1.0 visual acuity, and Fig.12 shows an image from the right eye with 0.5 visual acuity. Comparing these images, we can see that Fig.12 is affected by myopia more than Fig.11. Especially, it is seen that further object is more affected in Fig.12. The actual system displays both images simultaneously so that we can also understand influence from difference of the acuity very much. Figures 13 and 14 show images from different viewpoint with the same visual acuity, 1.0. From the images far objects is unclearly seen under healthy view of 1.0 acuity. 4.2 Evaluation For evaluation of the proposed system, questionnaire is sent out to 21 students in Chuo University.
12 among them have myopia and different visual acuity. After their experience on the CRT-based IPD, they return the questionnaire described as follows: 1. Can you understand difference of each visual acuity? 2. How do you think handleability when you set up visual acuity? 3. How do you think handleability when you translate view point? 4. Can you get depth in situation which objects wash out? 5. Do you expand an understanding of Myopia? They answer each question by grade 1-5. Here grade 1 is the best and 5 is the worst evaluation. Table 1 shows result of answers. Table 1: result of opinionaire 1 2 3 4 5 ques.1 (ques.1 by 12 with different acuity) 12 0 0 0 0 ques.1 (ques.1 by 9 with same acuity) 5 3 1 0 0 ques.2 2 4 10 5 0 ques.3 0 1 5 10 5 ques.4 5 9 5 2 0 ques.5 10 10 1 0 0 From the answer of the first question, it is seen that all examinees with different acuity understand the difference of acuity on the proposed system. Also most of examinees with the same acuity grade 1 or 2. Consequently, it is shown that the proposed system provides the difference sufficiently. Considering the answers of the fifth question, too, we can conclude that the proposed system can work well as an educational tool for teaching myopia. However, from the answers of the second and third questions, manipulation feature should be improved in the proposed system. Furthermore we should discuss suitable scene setting, since the result of the forth question shows not so good for depth effect. object in the given scene into account, we generate blurred images from clear image. The proposed system provides two images for both eyes so that the system provides stereoscopic view on IPD. In future paper, the followings should be considered: decrease of computation time improvement of interface improvement of interactivity Acknowledgement The authors would thank to the TISE Collaborative Research Project of Chuo University, KAK- ENHI, the Grantin-Aid for Scientific Research(C) No.19500092 from the Japan Society for the Promotion of Science (JSPS), and the Telecom Engineering Center. References [1] Internet website of Eye-athletic Research Institute : http://www.469up.com/ [2] Tadashi Oyama, Invitation to Visual Psychology and Approach to World of Vainglory, SAIENSU-SHA Co., Ltd, 2000. [3] Internet website of National Eye Institute : http://www.nei.nih.gov/ [4] Internet website of TAUCHI EYE CLINIC : http://www.tanouchi.com/index.html [5] Naoya Sagara, Mitsunori Makino, A Visual Simulation of Motion Blur in Consideration of Optical Flow, Proceedings of the 2001 International Technical Conference on Circuits / Systems, Computers and Comunications(ITC- CSCC2001), Vol.I, pp.470-473, 2001. [6] Internet website of Sangubashi Eye Clinic : http://www.sangubashi.com/english/index.html [7] Internet website of NIDEK Inc. : http://www.nidek.co.jp/gaiyou2.html 5 Conclusion This paper proposed a representation method of a scene on IPD, which is influenced by weak or different human vision, in order to teach myopic view as education. Taking visual acuity and distance of each
Figure 10: a clear image without focus Figure 11: an image with 1.0 acuity Figure 12: an image with 0.5 acuity Figure 13: an image with 1.0 acuity from near viewpoint Figure 14: an image with 1.0 acuity from far viewpoint