Freescale Tower system w/K70

PIC, Arduino, Raspberri Pi, discrete circuits - sometimes making it yourself is more fun
Post Reply
Adam
Posts: 2240
Joined: Wed Oct 23, 2013 9:50 pm

Freescale Tower system w/K70

Post by Adam »

So the computer we are using based on a Freescale K70 processor which is based on the ARM Cortex M4. The processor board can be combined with other I/O boards using interconnect boards to make a fairly complex computer system.

Main CPU board:
DSCN5491.JPG
Serial I/O board, also includes Ethernet:
DSCN5492.JPG
Storage expansion board. In addition to the compact flash module, there is an SD slot on the bottom:
DSCN5490.JPG
LCD board:
DSCN5493.JPG
Tower module. Attachment slots look suspiciously like PCI slots:
DSCN5494.JPG
The whole system assembled:
DSCN5496.JPG
There are also other boards you can get that work through the "flex bus" tower interface. Freescale sells this as a prototyping system for designing embedded systems. My project goal in this class is to implement a simple image processing system using the SD/USB interface to import a sample image, an edge detection algorithm to detect some edges and the LCD screen to display some results. We'll see how this goes.
kevm14
Posts: 15200
Joined: Wed Oct 23, 2013 10:28 pm

Re: Freescale Tower system w/K70

Post by kevm14 »

Enhance.
Adam
Posts: 2240
Joined: Wed Oct 23, 2013 9:50 pm

Re: Freescale Tower system w/K70

Post by Adam »

So the class is over. I was able to get the board to do some edge detection with some pretty good results. I ended up using the SD interface to both import and export the images since I could not get the LCD display working in time. Performance was terrible (not surprising) with the board taking 7-10 seconds to process one 512px image with single dimensional edge detection. Not quite good enough for real time performance. With some optimization, I could probably get this closer to a second, but still too slow.

I worked in BMP format for the project since it is a raster format w/o compression to make my life easier, but I have had to convert to JPG to post here.

Standard test image.
lena_g.jpg
lena_g.jpg (21.58 KiB) Viewed 815 times
lena_g.jpg
lena_g.jpg (28.36 KiB) Viewed 815 times
Nature scene.
mountain.jpg
mountain.jpg
Car.
racecar.jpg
racecar.jpg
The output is pretty good. Even with a 1-D edge detection, the human eye can still identify the image. The hard part is to use this shape to programmatically classify the object in the image further, which would take forever on this board. Still an interesting experiment, though.
Post Reply