Implement stream_mode for functions about dual-raspi-stereo-vision HOT 5 CLOSED

Implementation of stream_mode complete. But the image doesn't seem to undergo denoising when captured directly to a numpy array. This has resulted in reduction of trigger time on server side.
But socket transfer causes the overall trigger time to increase. This network latency can be reduced by increasing the socket.recv() buffer size.
So, tradeoff between trigger_time and image_quality.?

from dual-raspi-stereo-vision.

Find out a method of converting BytesIO stream into numpy array directly without much lag. This would help maintain image_quality while keeping the time almost same.
We want numpy array because we don't want to perform the time consuming task of reading/writing to disk.

from dual-raspi-stereo-vision.

Capturing image directly to ndarray removes the pink tint that is visible on normal-path-capture. Don't know why this is happening.

• Implemented the main process loop in stream_mode.

Remaining work:

• Find out the diff between direct-path and ndarray image capture
• Need to implement multiprocessing in the main process loop
• Testing and debugging yet to be performed.

from dual-raspi-stereo-vision.

• For the diff between direct-path and ndarray image capture, follow the following links:
  -- Link1: https://raspberrypi.stackexchange.com/questions/63428/different-image-quality-from-picamera-capture-to-numpy-array-and-path
  -- Link2: https://www.raspberrypi.org/forums/viewtopic.php?f=43&t=177570&p=1131893#p1131893

• Implementation of multiprocessing will be handled in the following issue: #5

from dual-raspi-stereo-vision.

Testing successful. Everything is working..!!

from dual-raspi-stereo-vision.