| Lost a pair of healthy eyes bring great trouble and inconvenience for the blinds in their daily life. At present, the number of blinds in the world is very large, according to the data from world health organization at 2010, there are about 8.248 million blinds in our country China, and the population of the low vision is about 67.274 million. In recent years, with the rapid development of computer software and hardware technology, wearable electronic products come into our lives, and bring a lot of convenience and fun for People‘s Daily life. All this give us the ability to bring some change for the lives of the vulnerable group blinds, share them the fruits of modern civilization, and enjoy the blessedness from the progress of science and technology.Existing navigation system tools for the blinds provide blinds and eyesight disabilities certain help with their daily travel, but there are more or less some problems of these tools in the process of actual use, such as expensive, poor experience, poor interactivity and real-time performance not strong. Based on this, this thesis plans to create a blind intelligent navigation glasses system which is humane, intelligent and popularization. According to the analysis of scene simulation that the blinds walking on the road in the open air, designing this intelligent navigation glasses system may include the functions such as safe obstacle avoidance, traffic state identification, identify and locate a zebra crossing, voice prompt, and human-computer interaction using, etc. By installing a small camera on the glasses, and use this camera take photos of the information of traffic light and zebra crossing on the road, identify the image information and make judgments by use the image recognition technology, and then conversion the results into the voice signal and prompt the blinds through the receiver finally. The blinds can "see" and identify objects to walk on the road safely by this way. In addition, in order to achieve safe obstacle avoidance in emergency, also use the ultrasonic technology in this system design, bring the second line of defense for the blinds.This blind navigation glasses system uses the modular structure design, by use the way of interface calling to the realize the transmission and processing of data, and the call of each function module. The identification algorithm of traffic state is mainly first use the Haar characteristics classifier to determine the position of the traffic lights in the image, then use the HSV color model to identify the color in this area. The identification and positioning of Zebra crossing algorithm is to detect the edge of an enhanced image and then extract lines in the edge figure. Also should set decision rules according to the characteristic of the zebra crossing, and use this rules to identify if the extract straight lines are all Zebra crossing.In this thesis, the realization of all the functions in the whole system are on the Raspberry Pi 2 type B development board based on the operating system of Raspbian. And use Python and OpenCV as system development environment, realize the image recognition part of the system by using the Python API of image processing and recognition function provided by OpenCV. Finally, execute the the realization test of all function module of the system and the system‘s performance through a simulation. Take photos of the actual road traffic scenes, and using this system to identification and detection. It can be seen from the test results that this system meet the functional requirements of set basically, and the recognition speed and accuracy are meet the expected, and the whole system has high stability, it also has large extensible space. |
