Che-Wei Lee

Training with Multimodal Virtual Reality

The pressure sensor is used to collect abdominal breathing signals, while the brain wave sensor collects brain wave changes during the process. In addition, a carbon fiber heater is installed to simulate the thermal sensation brought by long-term self-training. The Unity3D engine is used to create the system in the C# language. There is a human body model of his own face in the scene. The abdomen on the screen will change with the real breathing rate. When the breathing rate is slower, the user will observe thicker and redder lines on the clothes. According to the line, the user can adjust their breathing speed. Signal analysis and statistical analysis use Matlab and SPSS. The analysis is divided into two parts; device verification and signal analysis. Device verification is to compare the signal correlation coefficient between comparing equipment to verify the feasibility of the low-cost equipment. Signal analysis can be divided into brain waves and respiratory signals. The wave uses a fast Fourier transform to calculate the spectrum intensity and get the alpha and theta waves. The respiratory signal uses an averaging filter to eliminate excessive noise and peak detection to calculate the respiratory frequency.

Piano Learning System with Virtual Reality

The job description is game development. The bending sensor is used to determine whether the finger is bent, so as to simulate the action of playing the piano. The signal of bending is transmitted to the game script through the UART interface. Music notation creation is recording the time difference between each note and the selected note through the CSV writer. By reading the CSV file, the preset notes are displayed on the screen according to different times. There are different intervals of cubes (excellent, normal, and bad) that can generate corresponding scores according to the collision of the notes pressed at different times. In the 3D interface, the hand model will produce the visual effect of bending the fingers according to whether they are bent in reality. There are collision bodies at the front of the fingers, which correspond to the correct piano keys to produce the effect of pressing the keys.

Oral Wound Identification System

Converting the original wound image from RGB to HSV to reduce the impact of color and brightness on processing. The picture is converted to grayscale, and then set the threshold to determine the exact black and white area of the wound. I first use erosion to prevent the wound position from connecting with the white position, and expand to connect the white spots inside the wound. Then I use the outline to find the boundary and circle the wound location. The wound is filled with the area inside the contour with the pixels of the calculated white point. The total number of white point pixels is calculated. The intersection points with the control group are compared with the processed wound image. The result of the similarity is about 90%. 

P&P Micro Climate Station

This project is a collaboration with the Department of Architecture, National Cheng Kung University. When the data collection reaches 12 bytes, it will use the IoT transmission protocol (sigfox module) to send it to the database, and display it through the web and mobile APP, giving the user a visual presentation and real-time warning. I am responsible for the modular appearance concept design and data testing. The appearance refers to the design of the louver to increase the air intake to reduce the heat accumulation of the sensor, and the white 3D printing wire to avoid absorbing solar radiation. In addition, the sensor model reduces assembly, reduces the trouble of installing and replacing the sensor through the modular design.

Web music player

This project uses HTML to implement the web version of the music player to solve the problem that TSMC cannot use the external network to listen to music.

Web Music Player URL: https://c824751.github.io/web-player/
Github: https://github.com/c824751/web-player