| Visual perception provides humans with a rich and complex information set about their surroundings.People rely heavily on visual input to perform many important functions.The loss of vision will bring people a lot of inconvenience and danger.According to the statistics of the visually impaired people,which is collected by the World Health Organization(WHO)in 2010,the number of blind people in China has reached 8.248 million,and the number of people with low vision is up to 67.274 million,making China the country with the largest number of visually impaired people in the world.Some existing blind-assisted products provide some help for the blind in their daily life.However,problems such as high price,poor interaction,poor experience and low safety factor level make these products unable to meet more demands of the blind and cannot bring them better life experience.Based on this,this paper plans to develop a low cost,more practical,safe and portable Intelligent Blind Object Recognition APP System.With the analysis of blind people's living needs,the main function of the system includes object recognition,traffic light recognition,zebra crossing recognition,voice interaction and so on.Mobile phones and other mobile devices are designed as the carrier to obtain the surrounding image information around the blind with the camera on the device and transmit the data through wireless or 4G network.The background calls the interface to complete the data transmission and processing between various function modules and realize the various functions of the system.Object recognition uses image recognition technology based on YOLO v2,which can quickly and accurately identify objects around the blind and the number of identifiable objects is up to 9000,covering all types of objects that the people may touch.As for the traffic light recognition,the position of the traffic lights in the image is identified and judged with YOLO v2 at first,and then the status of traffic light in the traffic light area is determined in HSV color space.The recognition and location of zebra crossing are straight-line-extracted from the of the images which are processed after the processing of grayscale,equalization and filtering,and then the zebra crossing in the image is recognized and located according to the established conditions.Finally,the identified results which are converted to the form of voice broadcast are communicated to the user,realizing the happy vision for blind people to experience life through "watching".The functions of the whole system in this paper arc researched and developed with the Windows 10 operating system.Taking VS2015 and OpenCV as the development environment.C++library functions provided by OpenCV and YOLO v2 algorithm as the main development technologies,various functions of the system background are realized,and the design and development of the front-end APP are completed by using the tool of Android Studio.In addition,the voice service provided by IFLYTEK is utilized to realize the interaction function of the system.Finally,the function and performance of the system are tested by taking pictures of actual road scenes as samples.The test results show that all the functions of the system meet the requirements,and the recognition speed and accuracy are always maintained at a high level,showing a high stability,and has a expandable space at the same time. |
PDF Full Text Request