Mosaic View: Modest and Informative Display

Transcription

1 Mosaic View: Modest and Informative Display Kazuo Misue Department of Computer Science, Graduate School of Systems and Information Engineering, University of Tsukuba Tennoudai, Tsukuba, Japan Abstract Mosaic View is a mosaic picture composed of small pieces that contain images. The images used for the pieces can be arbitrary, so the overall image depicted in the mosaic can be independent of the images constituting the parts. The mosaic can display a modest overall image containing rich information using many small images. This article describe one method for making a mosaic, as well as several ways to update pieces in such a way that the order of placement of the pieces is clear to the observer. Two applications of the Mosaic View are also presented: the Modesty Mirror and the Forex Monitor. Keywords: Image Mosaic, Mosaic Picture, Peripheral Display, Ambient Information Visualization. 1 Introduction Display devices have become large and have high resolution. In addition, networks that convey information have become broadband. These improvements in technology enable us to see high fidelity images everywhere. However, we do not require high fidelity images everywhere. In some cases, we prefer modest representation, as exemplified by abstract paintings or ambient displays [4, 5]. We have been developing a style of representation that is both modest and informative. By intentionally using low fidelity and low resolution images, we hope to express rich information in a modest way. Finding a balance between providing information and presenting a modest overall image is challenging. Our basic idea is to create a mosaic (or image mosaic [2]) where each piece is a small image. The resulting image looks like a mosaic when seen from a distance, but shows more information when seen up close. We named this representation style "Mosaic View." This idea was inspired by the Wooden Mirror [8] and mosaic photographs 1. 2 Mosaic View The Wooden Mirror is a mechanical mirror made from 830 small pieces of wood. The mirror displays rough monotone images by controlling the angles of the pieces electronically to change their positions. This results in color differences using light and shade. The Wooden Mirror behaves like a mirror by displaying images taken by a camera in real time. The wooden pieces are Figure 1: Mosaic View consisting of pieces. 1 We can find many web pages about mosaic photographs using Google.

2 40 40 mm, which are as coarse as the pixels of a display device. Although the resolution is not very high, the mirror provides a moderate power of expression. and independent images in each piece and in the overall mosaic. For example, in image monitoring, the current image can be a mosaic composed of a series of past images. A mosaic photo is an image composed of many small photographs, which are put together to compose a larger overall image. If we look closely at a mosaic photograph, we can clearly see the small photographs that comprise it, but as we move away we can see the larger image that these small photographs compose. Mosaic photographs can achieve a higher resolution than that provided by the number of pieces by selecting photographs according to the color and shape desired for the overall image. 3 A Mosaic View is similar to a mosaic photo, but we do not need to select suitable images for the pieces. Images used as pieces are not used as they are. Their color tones are adjusted like those of the wooden chips to represent an overall image as a mosaic (see Figure 1). Mosaic View can be composed by arbitrarily arranging small images. We can display different 3.1 Making Mosaic View Outline of Making Mosaic View Figure 2 shows an outline of how the Mosaic View is created. An original image is changed to a mosaic using pieces that may or may not have the same image. The example in Figure 1 uses only one image for the pieces. Suppose that the number of pieces that compose a mosaic picture is n n. First, the original image is reduced to an image of n n pixels. The image might need to be made smooth. The image of n n pixels is called the reference image. Each pixel of the reference image corresponds to a piece of the mosaic picture, which determines a piece s tone. A piece image is reduced to the size of the allocated size of one pixel, and then the tone is changed to match the color of the corresponding pixel in the reference image where the piece will be placed. For example, if the corresponding pixel Figure 2: Outline of creating Mosaic View.

3 of the reference image is reddish, the piece image is changed to the red end of the spectrum. Finally, when all piece images are placed in their corresponding positions, Mosaic View is complete. 3.2 Changing Color Tone The basic idea is to modify the color tone of a piece image by referring to the corresponding reference pixel while preserving the color of the piece image. To modify the color tone, the color range must be reduced. We reduce both the color ranges of the piece images and the reference pixels. Then, we mix the colors of the piece images and the reference pixels. For example, we reduce the color ranges of the reference pixels and piece images to 1/2, and then add the R, G, and B values of the reference pixels to the R, G, and B values for every corresponding piece image. Generally, the color ranges of the reference pixels and the piece images do not need to be reduced equally. The ratio of reduction can be changed if necessary. Suppose that p, r, and p are a piece image, its reference pixel, and the piece image after the color tone has been modified, respectively. Moreover, suppose that p R, p G, and p B are the RGB values of piece image p. The modified RGB values of the same piece image, p R, p G and p B, are represented by the following expressions. p' R = ( s pr + t rr ) /( s + p' = ( s p + t r ) /( s p' G G G + B = ( s pb + t rb ) /( s + t) t) t) where s and t are parameters to control the reduction ratio of the color ranges of the reference pixels and the piece images, respectively. When s is larger than t, a wider range of color is allocated to the piece images than to the reference pixels. On the other hand, when t is larger than s, a wider range of color is allocated to the reference pixels than to the piece images. 4 Variations of Presentation 4.1 How to Arrange Pieces We can see the overall image of a mosaic from a distance, but we see many other images when we get close to it. For example, a mosaic consisting of pieces can present 576 small images, which all can be different from each other if desired. We do not need to worry about the arrangement of the piece images if we consider them to be purely a set of images, but there may be some reason for the positioning of individual piece images. We should consider the arrangement of piece images in certain situations. For example, suppose that we use a collection of photographs shot every 150 sec (2.5 min) as piece images. A mosaic could be comprised of images of the previous 24 hrs ( sec). In that case, the arrangement of the piece images is important to grasp them as a time series. With such a situation in mind, we devised a layout for piece images and a way to update them Horizontal- or Vertical-scanning Form As the names imply, the horizontal-scanning form arranges a series of piece images horizontally and the vertical-scanning form arranges them vertically. Figure 3(a) illustrates the arrangement of pieces using the horizontal-scanning form. The piece images are arranged from left to right. The rightmost position in one row is followed by the leftmost position in the next row. The numbers in Figure 3 show the order of the pieces. In the example described above, 24 photographs shot every 150 sec are arranged in a line. This means that one line corresponds to only 1 hr. The images become older as we look from left to right. We can see the newest image at any time by looking at the upper left corner, and we can see 1- hr increments by looking in the vertical direction Spiral-scanning Form The spiral-scanning form arranges a series of piece images in a spiral. Figure 3(b) shows an example of the spiral-scanning form. The piece images begin in the upper left corner and move in a clockwise direction as they spiral towards the center. An advantage of this form is that pieces with adjacent timestamps are always placed next to each other. In the horizontal-scanning form, pieces with adjacent timestamps might be placed far apart

4 from one another (see pieces -5 and -6 in Figure 3(a)). 4.2 Updating Pieces A time series can be expressed by arranging piece images in a horizontal-, vertical-, or spiralscanning form. If an observer does not know which form was used, the order of placement cannot be determined by observing the piece images. Even when the images change, if they all change at the same time, the order of the piece images still cannot be determined. For example, even if we know that the piece images are arranged using one of the three forms previously described, we probably cannot distinguish which one it is. Therefore, in addition to choosing a form of arrangement, we need a way to express a onedimensional order of the piece images. For example, when we use photographs shot at regular intervals as the piece images, we could add one new image at a time and discard the oldest image. We developed the following method as an expression technique of the time series of such images One-by-one Replacement One of the easiest techniques for changing the piece images is to replace them one by one, i.e., (a) Horizontal-scanning form the images are not all changed at the same time. First, the oldest image (X) is replaced by the second oldest image (X+1). Then, that image (X+1) is replaced by the third oldest image (X+2). The replacement continues one by one until the newest image is reached at the end. We naturally understand the order of the images because the images are replaced one at a time. The order would be difficult to recognize if the images all changed at the same time Shifting Pieces We can understand the arrangement of the piece images in the method above. However, understanding the positions to be renewed is difficult because each image changes instantaneously. Therefore, we developed another method of shifting pieces to new positions. First, the oldest image is removed, and then the second oldest image slides over to the newly vacant position. In other words, the movement of the piece image is animated. As each image is removed, the next piece slides to the older image s position. The movement continues one by one until the newest image is moved to the vacant position at the end. We can see the images move to their new positions one by one in this method. To prevent the piece images from jumping over each other, the spiralscanning form is preferable. This is because timeadjacent piece images are next to one another and easily slide over. The pieces move to the next vacant space one by one like the pieces of the 15 puzzles. The disadvantage of this method is that it requires a lot of time. A moderate amount of time is necessary to show smooth movement of each piece. For instance, in the example of the pieces mentioned above, if one piece moves at 0.1 sec, it takes approximately 1 min to renew all of the images. This is too much time for displaying moving images. 5 Examples We have developed two applications of the Mosaic View. (b) Spiral-scanning form Figure 3: Variations of moving pieces. 5.1 Modesty Mirror The Modesty Mirror displays a motion image taken with a USB camera to create a mosaic

5 picture (or mosaic movie). Each shot is arranged as a piece image. Obtaining Web cameras has become easy, and monitoring using web cameras is now done everywhere. However, excessive installations of Web cameras create a bad impression when people feel they are being observed. The Modesty Mirror softens this impression by displaying the images as a mosaic movie. Moreover, a certain past period can easily be observed by looking at the piece images, which is an advantage of this application. 5.2 Forex Monitor The Forex Monitor (foreign currency exchange monitor) displays changes in foreign exchange rates. A rising exchange rate is depicted with a cheery picture and a descending rate is depicted with a gloomy picture. The macro trend is expressed by the overall image of the mosaic while the micro changes in the rates are depicted by each piece image. For example, changes from the previous day s closing price in New York are displayed as an image of flowers when the rate has increased and as an image of cloudy skies when the rate has decreased. The change is expressed every 10 min in each piece image. A candlestick chart of every 10-min change can express information more directly than the Forex Monitor. However, the Forex Monitor can casually express such information for specific people. 6 Related Work Many ambient displays have been created in various shapes and forms. They use either physical or electronic displays to represent information in a graphic manner. Well-known examples of physical displays are water lamps and pinwheels [1]. Several researchers have investigated electronic displays [3, 4, 6]. Skog et al. used the term "ambient information visualization" for information visualization applications that reside in the environment or periphery of the user rather than on the screen of a desktop computer [9]. Redström et al. referred to ambient information visualization with artistic work as "informative art" [7]. Mosaic view is one method of creating ambient information visualization. The Forex Monitor described in Section 5 is one such application because the monitor shows the exchange rate as peripheral information. If the resulting mosaic (a) Modesty Mirror (b) Forex Monitor Mirror image is projected onto large plasma display as mosaic with past images for the pieces. Mosaic changes with exchange rate. Mosaic shows macro change, while each piece shows micro changes. Image of flower in front shows increase in rates. Image of cloudy skies in back shows decrease. Figure 4: Applications of Mosaic View.

6 image is artistic, the Forex Monitor is informative art. 7 Concluding Remarks We explained a technique called Mosaic View that can express rich information modestly. We imagined a situation where expressing information with high fidelity would be obstructive and created Mosaic View. We explained how to create Mosaic View and how to arrange the piece images to construct it. Mosaic View balances modesty and information. There are some factors that influence the balance. One factor is the number of piece images. Our future work will be controlling the balance by changing the number of pieces. 8 Acknowledgments The author thanks Mr. Takashi Nakamura of the Interactive Programming Laboratory at the University of Tsukuba for his help in developing the prototype. 9 References [1] Andrew Dahley, Craig Wisneski and Hiroshi Ishii, Water Lamp and Pinwheels: Ambient projection of digital information into architectural space, In CHI 98 conference summary on Human factors in computing system, pp , [2] Adam Finkelstein and Marisa Range, Image Mosaic, In Proceedings of 7th International Conference on Electronic Publishing (EP'98) held jointly with the 4th International Conference on Raster Imaging and Digital Typography (RIDT'98), Lecture Notes in Computer Science, Vol. 1375, pp , [3] James Fogarty, Jodi Forlizzi and Scott E. Hudson, Aesthetic information collages: generating decorative displays that contain information, In Proceedings of the 14th annual ACM symposium on User interface software and technology (UIST '01), pp , [4] Lars Erik Holmquist and Tobias Skig, Informative Art: Information Visualization in Everyday Environments, In Proceedings of the 1st international conference on Computer graphics and interactive techniques in Australasia and South East Asia (GRAPHITE 03), pp , [5] Hiroshi Ishii and Brygg Ullmer, Tangible Bits: Towards Seamless Interfaces between People, Bits and Atoms, In Proceedings of the SIGCHI conference on Human factors in computing systems (CHI 97), pp [6] Christopher Plaue, Todd Miller and John Stasko, Is a Picture Worth a Thousand Words? An Evaluation of Information Awareness Displays, In Proceedings of Graphics Interface 2004, pp , [7] Johan Redström, Tobias Skog and Lars Hallnäs, Informative Art: Using Amplified Artworks as Information Displays, In Proceedings of DARE 2000, pp , [8] Daniel Rozin, Wooden Mirror, 1999, [9] Tobias Skog, Sara Ljungblad and Lars Erik Holmquist, Between Aesthetics and Utility: Designing Ambient Information Visualizations, In Proceedings of IEEE Symposium on Information Visualization 2003 (InfoVis2003), pp , 2003.