How do I convert a csv file (16bit (high) color) to image in Python?

Question

Context: I built a small thermal camera which can save 70x70 pixels to an SD card. These pixels have a color value ranging from 0 to 2^16. (Actually certain colors, like black (value 0) are never displayed). This color is determined like explained here: c++ defined 16bit (high) color

I want to convert this data into an Image with my computer using Python.

An example gathered from another question unfortunately doesn't yield satisfying results: bad image

As you can see, the image doesn't look very good. My screen shows something like this (note that these two examples were not captured at the same time): photo

The white frame is not part of the csv-file.

This is the code I have used to generate the image: I have experimented with the color_max value but didn't get good results.

#Python CSV to Image converter
#pip2 install cImage
#pip2 install numpy

from PIL import Image, ImageDraw
from numpy import genfromtxt

color_max = 256
#original 256

g = open('IMAGE_25.TXT','r')
temp = genfromtxt(g, delimiter = ',')
im = Image.fromarray(temp).convert('RGB')
pix = im.load()
rows, cols = im.size
for x in range(cols):
    for y in range(rows):
        #print str(x) + " " + str(y)
        pix[x,y] = (int(temp[y,x] // color_max // color_max % color_max),int(temp[y,x] // color_max  % color_max),int(temp[y,x] % color_max))
im.save(g.name[0:-4] + '.jpeg')

This is the csv-file: Image Data

31 represents blue in this case, high values are more red.

Thanks for any help!

---

Here's some additional information about my project:

Arduino Thermal Camera with SD card support and image saving capability using the AMG8833 Thermal Imaging sensor made by Panasonic: Datasheet

GitHub (Arduino and Python Code)

3D-printable case used by me and original Arduino code

Schematic with SD card added

Answer 1

I think it should look like this:

#!/usr/local/bin/python3
import numpy as np
from PIL import Image

# Read 16-bit RGB565 image into array of uint16
with open('IMAGE_25.TXT','r') as f:
    rgb565array = np.genfromtxt(f, delimiter = ',').astype(np.uint16)

# Pick up image dimensions
h, w = rgb565array.shape

# Make a numpy array of matching shape, but allowing for 8-bit/channel for R, G and B
rgb888array = np.zeros([h,w,3], dtype=np.uint8)

for row in range(h):
    for col in range(w):
        # Pick up rgb565 value and split into rgb888
        rgb565 = rgb565array[row,col]
        r = ((rgb565 >> 11 ) & 0x1f ) << 3
        g = ((rgb565 >> 5  ) & 0x3f ) << 2
        b = ((rgb565       ) & 0x1f ) << 3
        # Populate result array
        rgb888array[row,col]=r,g,b

# Save result as PNG
Image.fromarray(rgb888array).save('result.png')