Haruki Kawanaka

Estimating heart rate using wrist-type Photoplethysmography and acceleration sensor while running

This study provides Heart Rate (HR) Estimation using wrist-type Photoplethysmogpraphy (PPG) sensor while the subject is running. We propose the algorithm to estimate heart rate for the wrist-type PPG sensor. Since body motion artifacts easily affect the arm portion, our method in this study also uses accelerometer built in the wrist-type sensor to improve the accuracy of heart rate estimation. Our method has two components. One is rejecting artifacts with the power spectrums difference between PPG and acceleration obtained by frequency analysis. The other is the reliability of heart rate estimation, defined by the acceleration. Experimental results while our test subjects were running came closer to the holter Electrocardiogram (ECG) in high accuracy (r = 0.98, SD = 8.7 bpm). We, therefore, report the heart rate estimation method which has a higher degree of usability compared to existing methods using ECG.

Driver's cognitive distraction detection using physiological features by the adaboost

Effects of drivers states adaptive driving support systems is highly expected for the prevention of traffic accidents. In order to create this constituent technology, detecting drivers psychosomatic states which occurs just before a traffic accident is essential. Therefore drivers distraction is thought as one of important factors. This study focused on detecting drivers cognitive distraction, a state which can easily lead to a traffic accident. We reproduced the cognitive distraction by imposing conversation or arithmetic loads to the subjects on a driving simulator. A stereo camera system were used as the means to track a subjects eyes, and head movements, which were set as classification features for pattern recognition on the Support Vector Machine (hereafter, SVM) basis used in the previous study of the AIDE project, a part of EU 6th Framework Programme. Diameter of pupil as well as the interval between heart R-waves (hereafter, heart rate RRI) from an ECG (electrocardiogram) were added for classification features to further improve the accuracy of drivers cognitive distraction detection. Based on this study, we established the methodology for more precise and faster drivers cognitive detection by using the AdaBoost.

Driver’s cognitive distraction detection using AdaBoost on pattern recognition basis

Detecting the mental and physical states which occur in a driver immediately before a traffic accident and then providing information to or warning the driver is an effective means of reducing traffic accidents. This study is focused on driver distraction, a state which can easily lead to traffic accidents, and reproduced this distraction in a driving simulator by providing conversation or arithmetic tasks to the subjects. Stereo cameras were used as the means to track subjectspsila eye and head movements. These movements were tracked and their standard deviations were set as classification features of pattern recognition, and the AdaBoost method was used to detect subject distraction. The interval between heart R-waves was also added as a classifier feature, in order to improve cognitive distraction detection performance. The results were then compared with the SVM method from the AIDE Project, which was carried out as part of the EU 6th Framework Programme.

Effect of pattern recognition features on detection for driver's cognitive distraction

Constituent technology of a driver monitor system using information of a drivers psychosomatic states is expected to create drivers states adaptive drive supporting system for the reduction of traffic accidents. In this study we identified a drivers distraction as one of major psychosomatic states which may result in a traffic accident by using Internet based survey on a questionnaire basis. Then we aimed at creating a methodology in use for detecting drivers cognitive distraction by means of using the AdaBoost which is capable of rapid and accurate classification. Furthermore we verified an effect of pattern recognition features such as interval between heart R-waves (hereafter, heart rate RRI) from an ECG (electrocardiogram), pupil diameter, and, gaze angle and head rotation angle.

Human Head Tracking in Three Dimensional Voxel Space

This paper proposes a new approach to track a human head in 3D voxel space. Information of both color and distance is obtained from multiple stereo cameras and integrated in 3D voxel space. Formulating a likelihood function from voxel location and its color information can achieve stable tracking with particle filtering in 3D voxel space

Robust background subtraction for quick illumination changes

This paper proposes a new method to extract moving objects from a color video sequence. The proposed method is robust to both noise and intensity changes in the observed image. A present background image is estimated by generating conversion tables from the original background image to the present image. Then, the moving object region is extracted by background subtraction. Using color gives more accurate detection than a previous method which used only monochrome data. Color images increase the computational load. The method addresses this problem by using the GPUs throughput. Results are demonstrated with experiments on real data.

Particle filter based tracking of moving object from image sequence

Object tracking is an important topic in computer vision and image recognition. The probabilistic approach using the particle filter has been recently used for the tracking of moving objects. Based on our trajectory recording system of the soccer scene with multiple video cameras at one view point, we propose the extended approach to increase the tracking robustness and accuracy using the particle filter. The proposed approach makes it possible to pass the necessary particle information using the color histogram and other key factors from one image to the next image, which are taken through the different camera scene with one PC. The performance of the proposed approach is evaluated in the experiments with real video sequence. It is shown that one PC can handle two video images in real-time.

Obtaining Shape from Scanning Electron Microscope using Hopfield Neural Network

In the environment of the Scanning Electron Microscope (SEM), it is necessary to establish the technology of recovering the 3D shape of a target object from the observed 2D shading image. SEM has the function to rotate the object stand to some extent. This paper uses this principle and proposes a new method to recover the object shape using two shading images taken during the rotation. The proposed method uses the optimization of the energy function using Hopfield neural network, which is based on the standard regularization theory. It is also important to give the initial vector that is close to the true optimal solution vector. Computer simulation evaluates the essential ability of the proposed method. Further, the real experiments for the SEM images are also demonstrated and discussed.

Cuffless blood pressure estimation using only photoplethysmography based on cardiovascular parameters

This study provides cuffless blood pressure estimation. In general, blood pressure changes when the subjects condition changes, and it is important to estimate it continuously and noninvasively. In many previous studies, they used PTT (Pulse Transmission Time) for estimating. However, PTT needs both electrocardiogram and photoplethysmography to be measured. Our method needs only a finger type photoplethysmographic sensor for estimating. We use the features obtained only from photoplethysmography for estimating, instead of PTT obtained from electrocardiogram. The features used are accelerated plethysmographys waveform, Heart Rate Variability and the rate of photoplethysmographys drift. Blood pressure is modeled as the product of CO (Cardiac Output) and TPR (Total Peripheral Resistance) in general. Then, we estimated blood pressure as the product of eCO and eTPR estimated by proposed photoplethysmographys features with Stepwise multiple regression analysis. Therefore, our proposed method provides not only blood pressure, but also CO and TPR. As of result, we estimated blood pressure based on eCO and eTPR, and we obtained r = 0.71. Therefore, we could obtain the result closer to Finometer in accuracy.

Extending Fast Marching Method under Point Light Source Illumination and Perspective Projection

An endoscope is a medical instrument that acquires images inside the human body. An endoscope carries its own light source. Classic shape-from-shading can be used to recover the 3-D shape of objects in view. Recent implementations have used the Fast Marching Method (FMM). Previous FMM approaches recover 3-D shape under assumptions of parallel light source illumination and orthographic projection. This paper extends the FMM approach to recover the 3-D shape under more realistic conditions of endoscopy, namely nearby point light source illumination and perspective projection. The new approach is demonstrated through experiment and is seen to improve performance.