Elphel Development Blog

10359 Stereo: Output Data Representation

So far, the code works in the following way:

1. Frames from sensor are stored in SDRAM, then the data go to the Cross-Correlation module.
2. The output of the C-C module are 256 values for each pass cross-correlation function. The address (=pixel displacement) of the maximal value is written to a BRAM.
3. After the whole picture line is processed the results are read out and form a line of a ‘depth frame’ that is sent to the main 353 board.

So, the ‘depth frame’ goes through the 353’s compressor and this makes the data available only for visual analysis. I’m planning:

1. 353 will get optionally ‘depth frame’ OR frames from the left/right sensor.
2. at the same time the ‘depth frame’ lines will be available through the I2C register – basically that means I’ll just duplicate the BRAM where I store the depth information.

Shall I keep the whole frame? shall I write it to the SDRAM? I’ll use 1 BRAM at first. Section 1 & 2 w/o ‘frames from the left/right sensor‘ will take less then an hour.