130 likes | 283 Views
Digital Image Processing with ImageNet Designer. Robotics II. General Concept: Graphical Programming. Use Blocks: No programming knowledge required Program a Block: programming knowledge required Modularization : Developers cannot influence each other with compile errors.
E N D
Digital Image Processing with ImageNet Designer Robotics II
General Concept:Graphical Programming • Use Blocks:No programming knowledge required • Program a Block: programming knowledge required • Modularization:Developers cannot influence each other with compile errors
Fast Development vs.Fast Execution • Fast Development: • MATLAB & Simulink with Video and Image Processing Blockset • Real-time fast Execution: • Hardware-near programming • OpenCV • Combination: • Develop in MATLAB and transfer to OpenCV • Or use ImageNet Designer
What is ImageNet Designer? • Developed at IAT (since Aug 2009) • Driven by the Visual Control of the FRIEND Robot • 3D Visualization of real Environment • Graphical programming • Directed graphs • Easy to use • Help included • Step-by-stepTutorials
Comparison OpenCV Results with ImageNet Designer Result • OpenCV: • Numerical Values • ImageNet Designer: • 3D Result Visualization • Offline Simulation Equivalent 4x4 Matrix, representing aCoordinate Transformation (also called Frame)
Why should we useImageNet Designer? • Execution Speed of OpenCV • but much simpler to use • No programming knowledge requiredto use OpenCV functions • Simulationof 3D-Results • Result in 3D is needed by our Robot (e.g. FRIEND) • Integrated Help • Video
Data types • Images (OpenCV IplImage*) • Color • Gray • Binary • Disparity • Matrix (OpenCV CvMat*) • Scalar (1x1 Matrix) • Histogram (256x1 Matrix) • 2D Objects • Points • Lines • Contours • Rectangles • Ellipses • 3D Objects • Points • Contours • Cuboids • Cylinder • Sphere • Coordinate Transformation
Window management • ImageNet Designer • Double click onpreview image • Zoom functionality • Pixel information • OpenCV • Hard codeshow window • Window will open on every execution(this can be annoying)
Histogram Equalization • Linearizes the Cumulative Histogram
Thresholding • ImageNet Designer • OpenCV IplImage* pSourceImage = cvLoadImage( “testimage.png”, 1 ); if( pSourceImage ){ IplImage* pGrayImage = cvCreateImage( cvGetSize( pInputImage ), IPL_DEPTH_8U, 1 ); cvCvtColor( pInputImage, pGrayImage, CV_BGR2GRAY ); //Convert the color image IplImage* pBinaryImage = cvCloneImage( pGrayImage ); cvThreshold( pGrayImage, pBinaryImage, 120, 255, CV_THRESH_BINARY ); cvNamedWindow( "window", CV_WINDOW_AUTOSIZE ); cvShowImage( "window", pBinaryImage ); cvWaitKey( 0 ); //Wait for keystroke }
Combination of Multiple Views • Join Images Block • Sums up arbitrarydata typesfor 3D-Simulation • See Video • Robot base
Where to downloadImageNet Designer • Project Page: • https://sourceforge.net/projects/imagenets • Support: • http://sourceforge.net/apps/mediawiki/imagenets • Student project: contact us! • lange@iat.uni-bremen.de • natarajan@iat.uni-bremen.de • enjarini@iat.uni-bremen.de