文字辨識相關

Posted on 週二 13 十月 2015 in Archive

Android 環境設定

  • 在AndroidStudio上用OpenCV SDK http://stackoverflow.com/questions/27406303/opencv-in-android-studio

  • Download latest OpenCV sdk for Android from OpenCV.org and decompress the zip file.

  • Import OpenCV to Android Studio, From File -> New -> Import Module, choose sdk/java folder in the unzipped opencv archive.
  • Update build.gradle under imported OpenCV module to update 4 fields to match your project build.gradle a) compileSdkVersion b) buildToolsVersion c) minSdkVersion and 4) targetSdkVersion.
  • Add module dependency by Application -> Module Settings, and select the Dependencies tab. Click + icon at bottom, choose Module Dependency and select the imported OpenCV module.
  • For Android Studio v1.2.2, to access to Module Settings : in the project view, right-click the dependent module -> Open Module Settings
  • Copy libs folder under sdk/native to Android Studio under app/src/main. In Android Studio, rename the copied libs directory to jniLibs and we are done. Step (6) is since Android studio expects native libs in app/src/main/jniLibs instead of older libs folder. For those new to Android OpenCV, don't miss below steps
include static{ System.loadLibrary("opencv_java"); }

(Note: for OpenCV version 3 at this step you should instead load the library opencv_java3.) For step(5), if you ignore any platform libs like x86, make sure your device/emulator is not on that platform.

  • Android Camera http://blog.csdn.net/yanzi1225627/article/details/8605061 http://ibuzzlog.blogspot.tw/2012/08/how-to-do-real-time-image-processing-in.html https://developer.qualcomm.com/software/snapdragon-sdk-android/facial-processing/user-guide

"最后提醒的是，如果程序中加入了previewCallback，在surfaceDestroy释放camera的时候，最好执行myCamera.setOneShotPreviewCallback(null); 或者myCamera.setPreviewCallback(null);中止这种回调，然后再释放camera更安全。"

  • Android camera & screen座標 http://blog.csdn.net/yanzi1225627/article/details/38098729

OpenCV Text detection

  • MSER http://www.mathworks.com/help/vision/examples/automatically-detect-and-recognize-text-in-natural-images.html#zmw57dd0e38 https://en.wikipedia.org/wiki/Maximally_stable_extremal_regions http://www.micc.unifi.it/delbimbo/wp-content/uploads/2011/03/slide_corso/A34%20MSER.pdf

不錯的中文解釋 http://blog.sciencenet.cn/blog-1327159-849648.html

  • Canny Edge Detector http://docs.opencv.org/doc/tutorials/imgproc/imgtrans/canny_detector/canny_detector.html

  • 把camera preview轉Mat, 灰階化 http://www.jayrambhia.com/blog/opencv-android-image/

Android onPreviewFrame傳回來的data是YUV4:2:0的形式

  • OpenCV image rotation http://opencv-code.com/quick-tips/how-to-rotate-image-in-opencv/

  • YUV to RGB http://stackoverflow.com/questions/25666120/converting-from-yuv-colour-space-to-rgb-using-opencv 一開始YUV Matrix的size不能亂設，不然轉換的時候會跑掉:P

OCR lib

  • Tesseract https://github.com/tesseract-ocr/tesseract