Machine View and Clouds Connectiing on Edison

by jamesnet44 in Craft > Digital Graphics

378 Views, 4 Favorites, 0 Comments

Machine View and Clouds Connectiing on Edison

Edison+opencv.jpg

Edison is powerful, which has a 500MHz Atom core and a 486 MCU @100MHz.

So I try to implement OpenCV on Edison, and do a 'shape' recogniazing. -- make different from normal ARM-cortex M.

Then send the result to Clouds. -- Edison has WiFi/BLE built in.

System Config

Edison_cam_con.jpg

1.OpenCV module should be installed in Edison's core image.

2. Edison's USB-A socket should be actived, work as a USB host. The uvc compatible camera will connected at this port. SW1 should be put at '2' , towards the USB-A socket. Use external DC power supply.

Tips: In this mode, Edison has only the WiFi channel to connect to ISS. You can connect Edison and PC to a same WiFI router if you want to debug your program and send data to remote clouds.

3. Intel System Studio is right choice for OpenCV application, comparing with XDK, Arduino.

Try OpenCV Function

camera_cap_9a.jpg

First thing is test the camera.

Using console command: lsusb, ls /dev/v* ... you can check if the uvc camera be connected.

And then you can test the camera by capture from it and save as a local file.

Tips: you need a tool like WinScp to check the file saved in Edison.

After checking camera, we can use OpenCV to identify simple shapes: rectangle, circle,....even if my camera gets bad focus.

My project does the 'rectangle' recognising, if you show a 'card like' paper in front of the camera, it will recognise it and send the result to the clouds.

Connect to Clouds

Edison_clouds_waveform.jpg
Screenshot_2016-10-29-17-04-55.png

According to Edison's document, I tried IBM bluemix, Intel analystics,,,,but failed.

In my area, it's not easy to connect google, facebook, youtube...the clouds server is too slow to do application.

So I changed to a local open/free clouds : OneNet. it's here: http://open.iot.10086.cn/

At there, you can create a Product, Device, and send data to the clouds, monitor your device on web explorer and app.

Conclusion

Edison_camera.jpg

Edison is powerful, and is easy to implement OpenCV on it.

It increases the capability of image processing in IOT devices.

By the strongly support of Intel specialist, I finished my first OpenCV project on Edison, and I believe there is a good future of Edion IOT module.