Task Description: Using the 3D printed Payload EM, take images of the sample plate from both camera locations then counting pixels of each sample.
Completed: 10 Aug. 2020
Conducted and documented by: Ali
Resources Required: EM payload module
Verification Activities: V-PLD-0190, V-PLD-0320, V-PLD-0330, V-PLD-0340
Pass Criteria:
Confirm that all samples are in view in both locations (V-PLD-0190)
Determine the number of red, blue, and green pixels for each sample from each camera location and confirm that:
Each sample is measured with no fewer than 59 red pixels (V-PLD-0320)
Each sample is measured with no fewer than 52 green pixels 52 (V-PLD-0330)
Each sample is measured with no fewer than 45 blue pixels 45 (V-PLD-0340)
The summary of total number of pixels, red, green and blue pixels are provided as follows.
According to Figure 1, it can be confirmed that all samples are in view in both locations (V-PLD-0190).
According to Table 1, it can be confirmed that all samples’ pixels number meet their minimum requirement which ensures that the color distribution matches the camera’s Bayer pattern.
Figure 1: Original image
In this test, the following assumptions and simplifications have been considered.
The payload will include two 2 MP ArduCam MT9D111 cameras located symmetrically with respect to the plate. Measurements will be taken with a single camera, with the second included for redundancy. However, in the test, the camera was installed in one of the positions and took the image.
The cameras’ positions are symmetric with respect to the sample plate. If we do another test by taking images with the second camera, the number of pixels for each well will be mirrored with respect to the sample plate’s centerline. Thus, doing another test for counting pixels with the camera in its second position would be unnecessary.
The camera is capable of outputting RAW format, has a known Bayer pattern, and has a sufficiently high signal-to-noise ratio and pixel count to provide statistically meaningful measurements. However, in this test, the images were taken in JPEG format instead of RAW.
The number of pixels in images with the JPEG format is over 6 times more than the number of required pixels for the furthest well.
We are using a wide-angle lens for the camera. Wide-angle lenses have a wide Field of View (FOV) and consequently, poor depths of field. So, everything is not in the focus of the camera. Thus, we expect a lot of blurring for the wells which are far from the focus of the camera. When some pixels are blurry, they contain surrounding pixels information instead of themselves. Therefore, in this test, blue colours used on yellow ones to see what areas appear green due to blurring [MD5] between the blue and yellow areas.
Apparently, there are a few tiny areas at the edges that seem blurry. Green areas are not obvious because of the white areas which are the subtract of paper due to cutting the holes. In total, blurry areas seemed small enough to be neglected and won't affect the result significantly.