Vol.2 No.4 2010

6/68

Research paper : Development of a real-time all-in-focus microscope (K. Ohba)−236−Synthesiology - English edition Vol.2 No.4 (2010) (b) First FS phaseTo incorporate the idea into an actual hardware system, high-speed variable focus mechanism and information processing and communication technology to accomplish the data acquisition, communication, and processing in real time were required to synthesize the system. For the high-speed variable focus mechanism that allowed dynamic observation at 30 frames/sec., response rate at 30 Hz was required, but available products only had single-digit response. However, this was solved when we visited a micromachine exhibition and came across the varifocal lens developed by the Denso Corporation in their micromachine project. For information processing and communication, when the number of necessary images to composite a single image was N, N × 30 frames/sec. × image digital size would be necessary. We then met Delft High Tech Corporation (current DHT Corporation) that imported the vision chip MAPP series from the Netherlands, and Kawatetsu Techno Research Corporation (current JFE Techno Research Corporation) that was working on its development. We were able to solve both technological issues and created Prototype 1.(c) Second FS phaseHowever, speed remained at 0.5 frames per sec., and the initial goal of 30 frames/sec. could not be achieved. We spent about a year negotiating with several companies to develop the high-speed vision chip. This was our first obstacle. In ordinary research, we could have written, “The problem can be solved by developing an ultra high-speed vision chip and implementing it to the device,” and then leave it. Yet, we set out on the voyage to clear the first obstacle from the desire to carry this project out to the end. However, even if we visited the companies with our request, they turned us down because they did not see any business merits for spending several hundred million yen for the development. The break came when we visited Photron Ltd. This was a high-speed camera manufacturer that had technology to shoot images at 10,000 frames/sec. and to transfer them to memory. They gave us encouraging advice that our requirement could be realized by a high-speed image capture device and the LVDS (low voltage differential signaling)Term 1 interface and by using the FPGA (field programmable gate array)Term 2 without developing the vision chip. Prototype 2 was created in a month.(d) Third FS phaseIn this phase, we started discussing future business with Photron. When we participated in exhibitions, we learned that the all-in-focus image was most highly demanded in microscope use, and therefore we started developments specializing in microscope application. Initially, the Denso varifocal lens was hand-made, mass production was not possible, and it was not suitable for high-precision positioning required for the microscope. We sought out the piezo-actuator of Physik Instrumente (PI) GmbH of Germany. In this device, the focal distance of the microscope was adjusted electrically, and we could obtain the desired precision by modifying the control unit. The second obstacle was to achieve high precision in high-speed mechanical movement, and the third FS phase was to overcome this issue. Initially, we were thinking only about the high-speed information processing, and expected the high-speed mechanical movement would somehow fall into place. In fact, we were plagued to the end on how to achieve highly reliable high-speed mechanical movement.(e) Product realization phaseBy repeating several demonstrations using the microscope system, we considered further business application by separating the uses into industrial and biological. However, the microscope system had a special sales channel where the optical manufacturer visited the researchers to sell it and then became in charge of maintenance. For an outsider like Photron to enter, it had to provide OEM to conventional optical companies and establish the sales channel to reach the users. This was our final obstacle. Over one year was needed to establish the sales channel, and the product was launched into the world in 2003.3 Idea phaseThe idea explained in the previous section was born from a simple scientific interest of “wanting to make an image with everything in focus.” The initial application of this function was vaguely considered for digital cameras and eyeglasses. We started by checking the efficacy by conducting offline processing on a PC for several images with different focal distances.In this phase, to conduct simple processing within the memory capacity and processing ability of the PC, we developed a sequential processing algorithm to increase the memory efficiency and a parallel processing algorithm to increase the processing capacity.3.1 Theoretical synthesis - depth from focus methodIt was described earlier that in case of optical microscope image, the issue of shallow depth of focus greatly affected the operability. At the same time, this may be a major advantage to realize the depth from the focus method[6][7] which is one of the 3D measurement methods of objects[5]. While the all-in-focus camera was a solution to the shallow depth of focus, by making use of this problem, not only could we achieve an all-in-focus image where every part was in focus, but also by using the depth from the focus method, it became possible to obtain the 3D shape of the object using only a single lens.Figure 3 shows the conceptual diagram of the depth from the focus method. When observing an object with different depths, the distance to the image surface is scanned, and the local frequencies of the dark-light data at each point of the

※このページを正しく表示するにはFlashPlayer9以上が必要です