How to change contrast and brightness of an image with Python (Specifically OpenMV)?

Mr.Meerkat · July 17, 2017

The problem:

The code for an robotic arm, specifically the part that senses (a camera) where the coffee is and dispenses hot water on top of it is currently having trouble detecting the correct object if there is any other round black object within its FOV or its unable to detect it at all if the 'coffee' is too far away.

Sample image of the picture taken by the camera:

Spoiler

So through photoshop, we can see that if we tweak contrast (and maybe brightness), we can remove/soften some of the surrounding features which helps the camera to detect the 'coffee' better.

Photo shopped images:

Spoiler

Contrast 100, stock brightness

Contrast 100, brightness 75

Currently the code is this (pastebin won't load in this country I'm on 'holiday' in [not Scotland] so...spoiler it is ):

Spoiler


import sensor, image, time


from pyb import UART


uart = UART(3,9600)


test = 'f'+'dfdfdfd'+str(12)


green_threshold   = (   0,   80,  -70,   -10,   -0,   30)


gray=(0,80)


coffee=(7,20, -2,15, 0,15)


region_min=30


region_max=200


black_ratio=0.5


num_white=3


white_threshold=115


deviation_length_width=20


sensor.reset()


sensor.set_pixformat(sensor.GRAYSCALE)


sensor.set_framesize(sensor.QQVGA)


sensor.skip_frames(10)


sensor.set_whitebal(False)


clock = time.clock()


data1=bytearray(6)


while(True):


	command_req=0


	out_xy=0


	if 0x55==uart.readchar() and 0x1A==uart.readchar() and 0x55==uart.readchar() and 0x1A==uart.readchar() and 0x01==uart.readchar()and 0x01==uart.readchar():


		command_req=1


	if command_req==1:


			out_buffer=chr(0x55)+chr(0x1A)+chr(0x55)+chr(0x1A)


			clock.tick()


			img = sensor.snapshot()


			blobs = img.find_blobs([gray])


			if blobs:


				 for b in blobs:


						 if b[2]>region_min and b[2]<region_max and b[3]>region_min and b[2]<region_max and b[4]>b[2]*b[3]*black_ratio:


							for i in range(1,num_white+1):


								if img.get_pixel((b[0]+i),(b[1]+i))<white_threshold :


									break


								else :


									 if i==num_white :


										 if abs(b[2]-b[3])<deviation_length_width:


											out_xy=1


											temp1=0


											temp2=0


											img.draw_rectangle(b[0:4])


											img.draw_cross(b[5], b[6])


											out_buffer+=chr(10)


											out_buffer+=chr(2)


											temp1=int(b[5]/128)


											temp2=b[5]-128*temp1


											out_buffer+=chr(temp1)


											out_buffer+=chr(temp2)


											temp1=int(b[6]/128)


											temp2=b[6]-128*temp1


											out_buffer+=chr(temp1)


											out_buffer+=chr(temp2)


											temp1=int(b[2]/128)


											temp2=b[2]-128*temp1


											out_buffer+=chr(temp1)


											out_buffer+=chr(temp2)


											temp1=int(b[3]/128)


											temp2=b[3]-128*temp1


											out_buffer+=chr(temp1)


											out_buffer+=chr(temp2)


											out_buffer+=chr(0)


											uart.write(out_buffer)


			if(out_xy==0) :


				out_buffer=out_buffer+chr(0x01)+chr(0x03)


				uart.write(out_buffer)

(Actual problem)

So I'm piss poor at Python (never 'properly' used it before but do have a really basic understanding of it) and online documentation isn't super helpful so here I am asking here on how you adjust the contrast and brightness of a picture in Python (specifically openMV) unless you know a easier and a less computational hungry method

Thanks guys!

On a side note, this is what it looks like when it is able to detect the 'coffee'

Spoiler

Azgoth 2 · July 17, 2017

How does OpenMV deal with images? Does it use something like a 2d array of pixel values? If so you could use some of the tools from probably the Numpy or Scipy libraries to clip your pixel values at a certain level (you'd probably need to play with the level a bit to find a good one). Or, if you could use the PIL library:

from PIL import Image, ImageEnhance
im = Image.Image("/path/to/saved/image")
contrast = ImageEnhance.Contrast(im)
contrast = contrast.enhance(FACTOR) # set FACTOR > 1 to enhance contrast, < 1 to decrease

# either save the image...
contrast.save("/path/to/new/location")

# or cast to a numpy array
import numpy as np
im_array = np.array(contrast.getdata())

Mr.Meerkat · July 18, 2017

8 hours ago, Azgoth 2 said:
How does OpenMV deal with images? Does it use something like a 2d array of pixel values? If so you could use some of the tools from probably the Numpy or Scipy libraries to clip your pixel values at a certain level (you'd probably need to play with the level a bit to find a good one). Or, if you could use the PIL library:
from PIL import Image, ImageEnhance
im = Image.Image("/path/to/saved/image")
contrast = ImageEnhance.Contrast(im)
contrast = contrast.enhance(FACTOR) # set FACTOR > 1 to enhance contrast, < 1 to decrease

# either save the image...
contrast.save("/path/to/new/location")

# or cast to a numpy array
import numpy as np
im_array = np.array(contrast.getdata())

you can't import libraries in OpenMV

Read a different section of the OpenMV docs and realised that I could just control the actual sensor instead (rather than editing the image afterwards)

However, found out it doesn't work as it still struggles detecting the correct thing and it did not increase the detection range )

Thanks anyway