Reconocimiento automático de formas

Abstract

In a very few years, the mobile phones have evolved at a vertiginous pace, starting as communication devices and turning into small computers ( Smart Phones) Android and iOs are ones of the most known operating systems for mobile terminals. In this project, we focused on Android platform, which is Open Source and uses the Java programming language. This project introduces the concepts of capturing, processing and image recognition. The aim of the plan is to make an able software to automatically recognise an image, or part of it, and show the closest image to it by comparing the original in a local database of the terminal. In order to make this possible, the project has been divided in two big points. The first one, was performed using the MATLAB, mathematical software tool. From an image stored on the user’s image gallery, and through a program created with the tool MATLAB, the five more similar images to the original one, were extracted from a database local computer hard drive. This process was done by the edge detection algorithm "Canny" and the "Bwdist", both of them calculate the distance of every pixel in the image to the nearest white pixel. At a later time we will be back to explain these algorithms and how the program runs. The second point of the project was to perform an Android app with the same goal as the first point. The difference is in the database, in this case, lies in the mobile SD card, so the image provided by the user may come from the phone's picture gallery or the camera of the terminal. This app uses calculations performed in Matlab and stored in a txt file, which in turn is stored in the phone's SD memory. This application may be a starting point for future extensions of the app, such as: become a client-server application; be able to expand the comparison data base by uploading photos in real time; allow internet connection; GPS location of the source image; etc.