Hiroshi Kage

Computer vision for interactive computer graphics

Vision can be a powerful interface device for computers because of its potential for sensing body position, head orientation, direction of gaze, pointing commands, and gestures. Such unencumbered interaction can make computers easier to use. We describe vision algorithms for interactive graphics and present vision-controlled graphics applications using these algorithms. Some applications employ an artificial retina chip for image detection or preprocessing.

Computer vision for computer interaction

Figure 1 shows a vision of the future from the 1939 Worlds Fair. The human-machine interface that was envisioned is wonderful. Both machines are equipped with cameras; the woman interacts with the machine using an intuitive gesture. That degree of naturalness is a goal today for researchers designing human-machine interfaces.

Data Processing for Multi-Channel Optical Recording : Action Potential Detection by Neural Network

Using a neural network, we have developed a program for fast and precise detection of action potentials (AP) in raw multi-channel optical recording data. The AP detection was performed in two steps: first, peaks were detected in raw optical data, and, second, the peaks were classified by the neural network into APs, noise and undecided peaks. The network was optimized and trained by the backpropagation learning algorithm, employing some thousands of manually classified peaks. The performance of the optimized network was found to be not completely satisfactory, although it was better than the classification by template matching and nearest-neighbor rules. The addition of a signal-to-noise ratio (SNR) of a peak to the network classification improved the classification performance: in comparison with the manual classification results, 96% of manually classified APs were detected. The causes of classification errors were discussed. In spite of the fact that the program required a slight amount of human intervention for undecided peaks, the program could allow mostly automatic AP detection.

Artificial retina large scale integration with on-sensor projection function for high-speed motion detection

We present an artificial retina (AR) large scale integration (LSI) with a high accuracy on-sensor image projection function. The fab- ricated chip has 32332 pixels, and each pixel is 56356 mm 2 large with a 41.5% fill factor. This AR LSI compresses 1024 bytes of input image data to 64 bytes of data by computing projection at the same time of image detection, on the focal plane. It also has a function of modulating the projection data into the form of spatial and temporal differentiation data, which fit a motion detection algorithm. Effectiveness of this function was proved by a motion detection experiment capturing an electric fan rotating from 775 to 1500 rpm speed. High sensitivity of 20 V/lx sec and wide dynamic range of 66 dB are also beneficial for high frame rates that lead to high-speed motion detection. Therefore, the AR LSI makes it possible to build a high-speed motion detection system even in combi- nation with an 8-bit microcontroller. This proves how effective it is to combine the processing function of the AR LSI and the recognition func- tion of the microcontroller to build up low cost image processing sys- tems.

Artificial retina chips as on-chip image processors and gesture-oriented interfaces

Players of a video game may sometimes find the use of conventional interfaces inappropriate. In such cases, we think that interfaces realized with a vision-based gesture recognition system may find favor. The artificial retina (AR) chip is a versatile image sensor whose use ranges from normal image acquisition to on-chip image processing, including on-chip image convolution. In this paper, we describe a gesture- input video game system, with the AR module including the AR chip, and motion-based gesture recognition algorithms. We showed that the algorithms can be accelerated by projection data, the direct output from the AR chip. To show its performance, we have applied our system to two commercially available video games.

Human intruder detection with Leaky Coaxial cables for wide area surveillance system

We propose an algorithm to detect human intruders by means of Leaky Coaxial cables (LCX) for wide area surveillance systems. When a human intrudes into an area between two LCX, the radiated waves will suffer considerable changes with specific fluctuations. The former version of our detection algorithm had many false detections because the algorithm used only amplitudes of in-phase/quadrature-phase (I/Q) detector signals. To develop a sophisticated detection algorithm, we analyzes I/Q detector signals in detail and introduce some critical signal features appeared in I/Q detector signals to discriminate human intruders from background noises and other irrelevant objects. We successfully reduced many of the false detections compared to the performance of the former algorithm. In this paper we introduce our LCX-based surveillance system, and describe a human intruder detection algorithm based on I/Q detector signals with typical simulation results. Current problems and remaining tasks are also discussed.

An Artificial Retina LSI with 5000 Frames per Second On-Sensor Projection Function

Our artificial retina (AR) LSI executes high-accuracy on-sensor image projection. The fabricated 32×32 pixel AR LSI compresses 1024 bytes image data to 64 bytes data, which reduces the image-processing load of the subsequent processing part. A motion detection system was developed through combination with an 8-bit microcontroller. This motion detection system outputted motion data at the rate of 50 reports/s for high-speed motion, corresponding to a 200 µs time difference. This high-speed image detection was attained also by the high sensitivity of 20 V/lxs and the wide dynamic range of 66 dB.