Development Environment

As a baseline example and starting point, Figure $[*]$ shows a prototype of a program that captures about 5 frames per second from an Android-enabled camera.

**Figure:** Example of an OpenCV-enabled application run on a low-resolution device with a keyboard (emulation). The number on the left represents the number of frames per second and the checkerboard pattern indicates that no camera is currently operating.

The interface is quite different for a high-resolution smartphone or a tablet, as shown in emulation mode in figures $[*]$ and $[*]$ .

**Figure:** Emulation of Android on a tablet-type device with OpenCV

**Figure:** Applications installed on Android on a tablet-type device with OpenCV

Figure $[*]$ shows the image obtained by applying Canny edge detector [5] to the synthetic scene shown before.

**Figure:** A simple program that applies the Canny edge detector to the video input (the default checkerboard pattern in this case) and the accompanying frame rate estimator on a dual-core AMD workstation (emulation mode). This may correspond to a high-resolution smartphone such as the Galaxy series or even a comparable tablet.

The SDK in use is Eclipse, which is further enhanced with Google addons, as shown in Figure $[*]$ . The program requires the Android SDK, NDK, typically Eclipse, and of course OpenCV (I used version 2.4.x).

**Figure:** Eclipse SDK and the two device types being tested with the simple program. Notice how the Eclipse toolbar is augmented by Google addons.

One would have to photograph the tablet to properly show this running with real data. In emulation I tested several device types, but the camera is emulated (the example in Figure $[*]$ is WVGA), unlike when it is run on the device.

**Figure:** The program in WVGA, emulated

For demonstration purposes I gathered some screenshots of the complete program running on a real device, a high-resolution tablet in this case. Figure $[*]$ shows some of those screenshots. These show the latest version of the program in ``About Impact'' mode - a menu option that enables tracking nearby obstacles. There are nearly 10 such menu options (the total varies depending on classification of distinction), but some are simpler filters.

**Figure:** Top left: car detection from a distance; Top right: detection from a shorter distance; Bottom left: car too close for detection; Bottom right: detection from afar. The program also comes with a menu and is not complicated to work with.