Tutorial on porting OpenCV(revision:5578) to android in Windows Environment

Software Prerequisites

• svn client (eg.Tortoise SVN)

·         cmake tool http://www.cmake.org/

·         Android NDK (revision r5 or newer) http://developer.android.com/sdk/ndk/index.html

·         OpenCV trunk https://code.ros.org/svn/opencv/trunk/opencv

     Additional requirements for Java wrapper and Android samples

·         SWIG http://www.swig.org/

·         JDK 5/JDK 6 http://www.oracle.com/technetwork/java/javase/downloads/index.html

·         Android SDK http://developer.android.com/sdk/index.html

·         ant http://ant.apache.org

     Windows additional prerequisites

At the moment OpenCV cross compilation under cygwin is not supported.

However you can cross-compile OpenCV on Windows using native port of make.
These two are tested to work for OpenCV:

·         http://gnuwin32.sourceforge.net/packages/make.htm

·         http://code.google.com/p/mingw-and-ndk/ (see "make-3.7z" under Downloads section)

Downloading OpenCV

Install the Tortoise SVN software on your machine, select a preferable directory and right click to check out.

·         Enter the URL with https://code.ros.org/svn/opencv/trunk/opencv and press ok

OpenCV build

By default OpenCV and the most of Android samples are configured for modern ARM-v7a architecture. 
If you want to compile OpenCV for emulator or older device, or want to build with NEON support please read the Android targets section for instructions.

     Windows (native)

Define the following environment variables:

• ANDROID_NDK - full path to the Android NDK. This path should not contain spaces, and tailing slash.
• CMAKE_EXE - full path to cmake executable (including executable name).
• MAKE_EXE - full path to native port of make (including executable name). This path should not contain spaces.

Instead of modifying your environment you can create file opencv\android\scripts\wincfg.cmd and put all settings to this file. OpenCV provides template for this file named opencv\android\scripts\wincmd.cfg.tmpl. Please note that settings from wincfg.cmd takes precedence over global environment variables.

Then open Windows console (cmd.exe) and go to opencv\android folder. Run script cmake_android.cmd from scripts folder:

cd opencv\android\

scripts\cmake_android.cmd

When script finishes you will have OpenCV compiled.

Troubleshooting

First, make sure that paths to OpenCV, Android NDK and make.exe do not contain spaces and other suspicious symbols.

If make fails with message similar to

*** target pattern contains no `%'.  Stop.

Then you probably use cygwin tools instead of their native ports. Remove cygwin from your PATH, delete build directory and rebuild.

Building console "Hello World" with OpenCV

This application is not usual Android application. This is a unix console which may be used as a template for development. Console application is much more convenient for development, debugging and performance optimization stages. Then, if you are finished with your CV algorithm, you can switch to normal Android development with Java + NDK (see the android-opencv and OpenCV_SAMPLE sections below).

     Prerequisities

First of all you need an Android device with root privileges. The goal is to have ability to run console applications on Android device using adb shell. Even chmod is not available by default (for example on Nexus One). You have the following options:

• Use emulator

·         Enable root access to your device. Use google to find instructions for your device.

• Use developer device, for example NVidia Tegra devkits have root access by default.

If the device allows you to do chmod and run executables, then follow the instructions below.

Second, you need OpenCV compiled for your device. By default this sample is configured for emulator. Read Android targets to know how to build OpenCV for emulator.

Also you need some free space on sdcard to run the HelloAndroid application - it will write small image to /mnt/sdcard. Please note, that even with emulator you need a sdcard. You can create it with AVD Manager from Android SDK.

     Windows

Set the same environment variables as needed for OpenCV build (variables set in \opencv\android\scripts\wincfg.cmd work as well).
If you are using wincfg.cmd file then ensure that ARM_TARGET and BUILD_DIR are set to correct values. Use the following values for emulator:

SET BUILD_DIR=build_armeabi

SET ARM_TARGET=armeabi

When all variables are set run the commands

cd opencv\android\apps\HelloAndroid

cmake_android.cmd

open CMakeLists.txt and find line :

SET( OpenCV_DIR ${CMAKE_SOURCE_DIR}/../../build )

Delete the cache in this line

Connect to your device with adb tool from Android SDK and run the application

Edit run.cmd line:

%adb% push .\bin\%OUT_DIR%\%PROJECT_NAME% /data/bin/sample/%PROJECT_NAME% || GOTO end

to

%adb% push .\build_armeabi\%PROJECT_NAME% /data/bin/sample/%PROJECT_NAME% || GOTO end

Next, go to opencv\android\apps\HelloAndroid\build_armeabi directory and change the opencv filename to HelloAndroid

Now execute the following command in command prompt:

run.cmd

If you can see HelloAndroid.png in your current directory and Hello Android! message in console - you've arrived.

Troubleshooting

If you see error message similar to

ld: cannot find -lopencv_contrib

or (on Windows)

ld.exe: cannot find -lopencv_contrib230

It means that cmake failed to find OpenCV compiled for Android. Check that all paths are set to correct values, remove build directory and try to rebuild the application.

Building the android-opencv project

The android-opencv project is a shared OpenCV library for Android. It contains some java bindings and a java/jni camera client. It is required to build the most of Android samples.

     Prerequisities

You need OpenCV compiled for Android.

     Windows

In addition to variables set for OpenCV build set the following environment variables (or put them to opencv\android\scripts\wincfg.cmd)

• ANDROID_SDK - full path to Android SDK.
• SWIG_DIR - full path to SWIG.
• ANT_DIR - full path to Apache Ant.
• JAVA_HOME - full path to JDK. Both x86 and x64 work for OpenCV. But note that Android SDK installer requires 32 bit JRE. At the same time if you download zip with SDK, you can work with 64 bit JRE/JDK without 32 bit version installed.

All paths should be specified without tailing slash.

Next go to android-opencv folder and run build script:

cd android-opencv

cmake_andoid.cmd

Troubleshooting

If android helper tool complains (when you run project_create.sh script on linux/Mac or cmake_andoid.cmd on Windows) then you have not installed API level 7 package to your Android SDK. You can either install it or update project settings to use one of API levels you have.

Put your target API level to default.properties file and rerun the script. android-opnecv supports target-7 or higher.

To get full list of installed API targets you can use the command

android list targets

Building a sample OpenCV_SAMPLE

     Windows

Go to OpenCV_SAMPLE directory and execute the build and cmake:

cd OpenCV_SAMPLE

..\..\scripts\build.cmd

cmake_andoid.cmd

Recreate build.xml file for the project:

set path=%PATH%;C:\dev\android-sdk-windows\tools 

android update project --name OpenCV_SAMPLE --path .

Plug in you device to compile and install the project:

path=%PATH%;C:\dev\apache-ant-1.8.2\bin

ant clean

ant compile

ant install

Screenshot of my computer with canny filter:

     Android hardware targets

You may wish to build OpenCV and samples for multiple hardware targets.

To do so, edit the name of file wincfg.cmd.tmpl to wincfg.cmd. Next open the file and Uncomment the line for the chosen device.

Just change the cmake cache ARM_TARGET to either: "armeabi" "armeab-v7a" "armeab-v7a with NEON"

You may install each of these to the toolchain/NDK, and they should be linked against properly via way of the android-cmake toolchain.

     Android API levels

You may wish to build OpenCV and samples for different level of Android native API.

Just change the cmake cache ANDROID_API_LEVEL to appropriative value.

The following levels are currently supported:

Platform Version	SDK API Level	NDK API Level
Android 3.1	12	9
Android 3.0	11
Android 2.3.4	10
Android 2.3.3
Android 2.3	9
Android 2.2	8	8
Android 2.1	7	5
Android 2.0.1	6
Android 2.0	5
Android 1.6	4	4
Android 1.5	3	3
Android 1.1	2	-
Android 1.0	1

If there are any doubts or hesitation or suggestion, feel free to write to jasongoh1987@gmail.com and i will update the blog from time to time. Thank You.

Taken and edited from http://opencv.willowgarage.com/

Credits to Andrey Kamaev