Several selected student projects are presented. These tasks are used as final year end test projects for undergraduate students in Soonchunhyang University, South Korea. To be able to use the machine vision concepts learned from this book, we encourage readers to choose at least one team-based student project. The project may be selected from this chapter or self-selected by students. For better evaluations of the suggested projects, screen captures of the student reports are included here. LabVIEW block diagrams are not included in this section. There can be a lot of different approaches to implement the projects and creative programing is always recommended!
Project 1: Noncontact Motion Measurement and Its Analysis
As seen in
Figure 19.1, an object is in pendulum motion. The motion is to be measured and analyzed in real time via image analysis. The trajectory of the object is plotted on graph. Also, the overlay on the object in the image display can be used for motion measurement verification. As an example, the period of pendulum motion ( T) can be calculated from the measured motion. From the period T, calculate the gravity constant g to confirm that g is about 9.81 m/s at the sea level. Note that the pendulum motion is one of the examples for noncontact motion measurement. You can extend this concept to other noncontact motion measurement applications.
Hint: Color pattern matching, pattern matching, tracking method)
Project 2: Intelligent Surveillance Camera
Develop a low-cost and multifunction surveillance camera system by using a low-cost USB camera (
Figure 19.2). In the case of conventional CCTV (closed circuit television), all the images are being recorded. To investigate and find any unusual occurrences, all the images need to be searched, which may take a significant time. So, unusual occurrences (e.g., people coming in through a door) need to be detected in real time. Then, an image or AVI file of the occurrence is saved. In this way, the use of disk memory for file saving can be minimized. Also, the searching time for events can be reduced significantly. After saving the image files (or AVI), the monitored results (image or AVI files) can be sent to security personnel via e-mail. At the same time, a text message can be sent to the securities via a mobile phone.
Program concept: An image is acquired via USB camera. A template image is saved as a reference image for (color) pattern matching. If the acquired images are different from the reference template image, the acquired images are saved for a specific time period or until they match with the template image. (The image can differ from the template image when the door is open and somebody comes in through the door.)
Hint: (Color) Pattern matching, Image save, AVI write)
Project 3: Driving a LEGO NXT Car (LEGO Mindstorms) with Finger Motion
Edge detection techniques can find the number of fingers along the ROI line (see
Figure 19.3). Use fingers along a ROI line to control remote control device. For example, a remote control car (via Bluetooth) made by Lego NXT can be used. Two different line ROIs may be used to count the number of fingers. If there is no finger in both line ROIs, the remote control car does not move. If there are fingers in the ROI in the left (or right), the remote car makes left (or right) turn. If there are fingers in both ROIs, you can make the remote control car go straight ahead (see Figure 19.4).
Hint: Multiple ROI, edge detection methods)
Project 4: Piano Keyboard Using Machine Vision
Multiple ROIs can be used as a noncontacting switch or selecting a specific function without actually touching a switch or device. As an example, develop a virtual piano keyboard using multiple ROIs, which will divide the image into many selectable areas.
Figure 19.5 shows five different selectable keys. Note that you may make more selections just by making more ROIs. In order to make multiple ROIs, GetImageSize is used to obtain the total pixel size of the image. Then, the image is divided by the number of ROIs to define any selectable areas. If you select any ROIs by using finger motion, the existence of fingers in ROIs can be detected by using the particle analysis ( Figure 19.6). When the fingers are detected, generate sounds that correspond to the keyboard action.
Hint: Multiple ROI, binary image, particle analysis)
Pendulum motion measurement. Figure 19.1
Intelligent surveillance system. Figure 19.2
Edge detection techniques for finding the number of fingers. Figure 19.3
Driving a remote control car with figure motion only. Figure 19.4
Piano keyboards using machine vision. Figure 19.5
Binary image conversion (Particle analysis). Figure 19.6