Hideki Komagata

Indoor positioning using a high-speed, fish-eye lens-equipped camera in Visible Light Communication

We propose a method for determining the indoor position based on a Visible Light Communication (VLC) system that uses a high-speed fish-eye lens-equipped camera. In VLC, lights are used as data transmitters and users can receive location information using a receiving device. Since the lights are configured to meet a pre-determined illumination level, the system requires neither space nor additional power. The lights serve their usual function as sources of illumination, and as a means of transmitting digitized information. Up to now, photo sensors or a normal lens-equipped camera have been used as receiving devices. However, the number of lights that can be received by a photo sensor is limited, and the receiving angle of the image that a normal lens-equipped camera takes is also limited. Since a camera with a fish-eye lens commands a 180-degree view of the ceiling, the number of detectable LED lights increases and positioning accuracy is improved. In terms of data transmission in VLC, the LED lights transmit data at 9.6 kbps. Thus, we use a high-speed complementary metal-oxide-semiconductor image sensor to receive the digitized information. The sampling frequency of the image sensor is up to 48 kHz. The LED lights send ID frames containing a prefix, ID, and Cyclic Redundancy Check (CRC) code. The ID and CRC are modulated with 4 Pulse-Position Modulation. The receiver detects the LED lights from the ceiling image. Then, variations in intensity at the center of the LED lights are stored as light signals. Since received data can be separated into two segments within a small buffer, according to the measurement period, received data are occasionally swapped. The receiver obtains the world coordinates of the LED light from the received ID. Finally, self-location estimation is performed using the relationship between the fish-eye image coordinates and the world coordinates. We conducted an experiment using the VLC platform in Niigata University, with the receiver position fixed at 24 measuring points. The platform area measures 5.4 m by 7.5 m, and the ceiling height is 3 m.The results show that the maximum horizontal error is 10 cm.We conclude that it is possible to determine a horizontal position within merely 10 cm, using the proposed method.

LED-tracking and ID-estimation for indoor positioning using visible light communication

We are focusing our research on indoor positioning technology; specifically, a type that uses Visible Light Communication (VLC); modulatable LED lights transmit data at 9600 bps, using 4 Pulse Position Modulation (4PPM), while a fish-eye lens-equipped camera receives the light signal over a 160-degree field-of-view. This type of lighting requires neither additional space nor-power. We assigned a unique ID to each LED, in order to recognize its position. Self-location is calculated from the relationship between the LED positions and coordinates on the image plane. In our previous research, we confirmed that self-location can be determined within 10 cm, using our system. However, we needed to attach dedicated transmitters to each LED used for positioning, especially in large buildings such as hospitals and shopping malls. So, in this paper, we propose LED-tracking and ID-estimation using LEDs with known IDs; doing so will significantly reduce the cost of installing- and running transmitters. Additionally, with the increased use of LEDs for positioning, accuracy naturally improves. We conducted experiments with the camera moving in 2 different environments: a) a small area, with just 4 LEDs; b) the VLC platform with a total of 24 LEDs, to demonstrate that as many as 13 LEDs can be identified. With 2 or more IDs detected beforehand, unidentified LEDs, as well as some that failed to be tracked, could be estimated, while the camera was in motion. Average positioning error in the smaller environment and the VLC platform were 3.78 cm and 6.96 cm, respectively. From this, location can be determined, even when some LEDs are offline.

An accurate method of extracting fat droplets in liver images for quantitative evaluation

The steatosis in liver pathological tissue images is a promising indicator of nonalcoholic fatty liver disease (NAFLD) and the possible risk of hepatocellular carcinoma (HCC). The resulting values are also important for ensuring the automatic and accurate classification of HCC images, because the existence of many fat droplets is likely to create errors in quantifying the morphological features used in the process. In this study we propose a method that can automatically detect, and exclude regions with many fat droplets by using the feature values of colors, shapes and the arrangement of cell nuclei. We implement the method and confirm that it can accurately detect fat droplets and quantify the fat droplet ratio of actual images. This investigation also clarifies the effective characteristics that contribute to accurate detection.

Multispectral filter array and demosaicking for pathological images

A single-shot multispectral camera equipped with a filter array has the potential to promote a fast and low-cost multispectral imaging system. We focus on the design of a multispectral filter array and demosaicking in this paper and propose a pathology-specific multispectral imaging system. The spectral sensitivities and patterns of the filter array are optimized by using training data of real pathological tissues. The mosaicked image is demosaicked by considering the designed filter array. We show the effectiveness of the proposed imaging system by comparing the recovered spectrum and RGB image with conventional methods.

An algorithm to evaluate the number of trabecular cell layers using nucleus arrangement applied to hepatocellular carcinoma

Recent advances in information technology have improved pathological virtual-slide technology and diagnostic support system studies of pathological images. Diagnostic support systems utilize quantitative indices determined by image processing. In previous studies on diagnostic support systems, carcinomatous areas of breast or lung have been recognized by the feature quantities of nuclear sizes, complexities, and internuclear distances based on graph theory, among other features. Improving recognition accuracy is important for the addition of new feature quantities. We focused on hepatocellular carcinoma (HCC) and investigated new feature quantities of histological images of HCC. One of the most important histological features of HCC is the trabecular pattern. For diagnosing cancer, it is important to recognize the tumor cell trabeculae. We propose a new algorithm for calculating the number of cell layers in histological images of HCC in tissue sections stained by hematoxylin and eosin. For the calculation, we used a Delaunay diagram that was based on the median points of nuclei, deleted the sinusoid and fat droplet regions from the Delaunay diagram, and counted the Delaunay lines while applying a thinning algorithm. Moreover, we experimented with the calculation of the number of cell layers with our method for different histological grades of HCC. The number of cell layers discriminated tumor differentiations and Edmondson grades; therefore, our algorithm may serve as an index of HCC for diagnostic support systems.

Feature analysis of cell nuclear chromatin distribution in support of cervical cytology

Abstract. Cytology, a method of estimating cancer or cellular atypia from microscopic images of scraped specimens, is used according to the pathologist’s experience to diagnose cases based on the degree of structural changes and atypia. Several methods of cell feature quantification, including nuclear size, nuclear shape, cytoplasm size, and chromatin texture, have been studied. We focus on chromatin distribution in the cell nucleus and propose new feature values that indicate the chromatin complexity, spreading, and bias, including convex hull ratio on multiple binary images, intensity distribution from the gravity center, and tangential component intensity and texture biases. The characteristics and cellular classification accuracies of the proposed features were verified through experiments using cervical smear samples, for which clear nuclear morphologic diagnostic criteria are available. In this experiment, we also used a stepwise support vector machine to create a machine learning model and a cross-validation algorithm with which to derive identification accuracy. Our results demonstrate the effectiveness of our proposed feature values.

Demosaicking Method for Multispectral Images Based on Spatial Gradient and Inter-channel Correlation

Multispectral images have been studied in various fields such as remote sensing and sugar content prediction in fruits. One of the systems that captures multispectral images uses a multispectral filter array based on a color filter array. In this system, demosaicking processing is required because the captured multispectral images are mosaicked. However, demosaicking is more difficult for multispectral images than for RGB images owing to the low density between the observed pixels in multispectral images. Therefore, we propose a demosaicking method for multispectral images based on spatial gradient and inter-channel correlation. Experimental results demonstrate that our proposed method outperforms the existing methods and is effective.