omiio.org is dedicated to fostering open source growth for mobile applications software development on TI’s OMAP™ platform.
Through omiio.org, you can collaborate, share and grow open source hardware and software.
In recent age, due to many reasons, the visual quality of an individual gets affected and maybe a loss of eyesight too. Congenitally visually impaired individuals face the visual problem along with other individuals with lower level visual problems like color blindness. Using Panda Board, Linux platform and OpenCV libraries, we have come up with the idea of video, image and sound processing techniques to aid the visually impaired people. Using programs and interfacing them with the hardware like the Panda Board and web-cam, and using real time based video processing, providing information to the user about the objects coming up with sound output. The platform used is Linux with libraries included from OpenCV and programmed using MS Visual Studio. The application is in locating everyday items using Image-Video Processing, Face Detection and Tagging, finding the distance of any obstacle from the present place and provide output as sound. The approach would go as taking a sequence of images in real time, enhance it with noise removal and other processes, detect the object or face in the video. Then, searching in the given database for pre-built objects or faces and provide the output in terms of names for the faces if present or the object identification. If not present in the database, then generate an audio signal for the same. Face tagging and detection along with object recognition programs are freely available over OpenCV websites which are further improved to suit our project. Interfacing web-cam with Panda Board with Linux platform and execute the program. Using Panda Board provides an efficient real time response to the input video and executes the programs of detecting and tagging the objects. Panda Board has specifications of 1.2 GHz processor with 256 MB RAM and external storage space can be provided. USB and Gigabit Ethernet ports are provided which can be used to interface keyboard, mouse or web-cam. HDMI port can be used to initiate the Panda Board with the executable files. Bluetooth and Wi-Fi ports are used for further extension. Web-cam, if stereo, would provide better resolution and fps in order to function in real time with no time lag. Human eye fps is around 30 and using Visual Studio with a well designed web-cam, fps of around 25 can be achieved which can function well in real time. Our main focus has remained on increasing the fps with less time lag between actual time and execution time. Stereo cameras are used to increase the resolution as well as the depth measuring accuracy. Observation that were made are the time difference, though in fraction of seconds, has to become equal. Image prediction technique can be applied to reduce time lag. Audio output is to be made accurate in order to help the user with real time process. The technology can be applied to medicinal science for human eye replacement as well as helping visually impaired to 'view through sound'. This can be applied to higher level applications in terms of scientific or industrial levels like planet and its distance detection in astronomical science, traffic management et alli.