4. Tactile pavement detection system using image recognition

Figure 2 shows the system configuration between personal computer with MATLAB, web camera, Arduino microcontroller board, XBee transceiver, voice module WTS 020, and speaker in order to give auditory warning to visually impaired people after the implementation of vision-based tactile detection method. After a coding has been inserted into the Arduino microcontroller, it will be ready to receive signals from MATLAB, and then send commands to the voice module to play selected audio files. In order to produce auditory output, a voice module will be used to play the required audio file when commands are executed. Figure 3 shows the actual hardware, which has been developed in order to validate the performances of the proposed vision-based tactile pavement detection system.

From the illustration, which is shown in Figure 3, a web camera is mounted on the center of the electronic cane. A distance between the webcam to the tactile paving is about 50 cm. The web camera is connected to the personal computer, which has been installed with the MATLAB software through XBee wireless communication. The personal computer will process the image, which has been captured through the web camera by using the proposed tactile pavement detection system. After the shape of the tactile pavement has been successfully determined by the proposed tactile pavement detection system, two types of voice guide will be given, which are WARNING and DIRECTION, through user's Bleutooth headphone. The result of the detection will be sent through the XBee transceiver to guide the cane's transceiver in order to activate the voice module. The voice guidance will be given through Bluetooth wireless communication.

Figure 2. System configuration for vision-based tactile paving detection system.

Vision-Based Tactile Paving Detection Method in Navigation Systems for Visually Impaired Persons 35 http://dx.doi.org/10.5772/intechopen.79886

Figure 3. Developed blind navigation system hardware.

This vision-based system consists of five main phases. The first part is to input the image containing the tactile pavement with the warning tactile and directional tactile. The second part is the preprocessing of the input images, which includes the filtering of the noises for the tactile detection in the image. The third part is to extract and determine the area and perimeter of the connected components detected in the image. The fourth part is to determine the metric for the connected components by using the area calculation algorithm of the detected components. The last part is to produce the accurate audio output to the visually impaired people. A process flowchart regarding the overall process of this system is shown in Figure 4.

#### 4.1. Input image

A webcam/camera will be used to capture the image that contains the pattern of tactile pavement. It will be loaded into MATLAB for further preprocessing to successfully detect any possible tactile shapes. Figure 5(a) and (b) shows the images of tactile paving, which are warning tactile and directional tactile.

#### 4.2. Preprocessing

This phase of the whole process is to filter the image actually required to be detected in MATLAB. Several steps are shown in Figure 4 in preprocessing, which are important to achieve the goal of the vision-based system.

Figure 4. Overall process flowchart for vision-based tactile pavement detection system.

Figure 5. Input tactile image. (a) Warning tactile and (b) direction tactile.
