PR: Simulation of CVD - Machado (2009)

[KelSolaar]

References #64.

[landscape-bot]

Repository health increased by 0.01% when pulling eec76cb on feature/cvd into 66d797b on develop.

• 10 new problems were found (including 0 errors and 8 code smells).
• 1 problem was fixed (including 1 error and 0 code smells).

[KelSolaar]

I would like to find a better name for that definition, it basically does almost the same job than colour.anomalous_trichromacy_matrix_Machado2010 definition except that it is using the pre-computed matrices from colour.blindness.dataset.machado2010 module.

[coveralls]

Coverage decreased (-0.01%) to 97.983% when pulling 9f7e3f7557c83c5c33e12b342a46097ab0bf51b5 on feature/cvd into 3933191 on develop.

[coveralls]

Coverage decreased (-0.01%) to 97.983% when pulling 835169e0cc367597e6506d4ba333edc88ed5f730 on feature/cvd into 3933191 on develop.

[coveralls]

Coverage increased (+0.01%) to 97.776% when pulling dca90c2 on feature/cvd into 0df6ebe on develop.

[KelSolaar]

@MichaelMauderer : So looking at the pre-generated matrices and the shift we induce for Tritanomaly, it seems to make sense, e.g. a 59nm shift as suggested by the highest value pre-generated matrix:

import numpy as np

import colour
from colour.plotting import *

colour_plotting_defaults()

cmfs = colour.LMS_CMFS['Smith & Pokorny 1975 Normal Trichromats']
cmfs_cvd = colour.anomalous_trichromacy_cmfs_Machado2009(cmfs, np.array([0, 0, 59]))
colour.CMFS[cmfs_cvd.name] = cmfs_cvd
multi_cmfs_plot(['Smith & Pokorny 1975 Normal Trichromats', cmfs_cvd.name], legend_location='lower right')

In the light of this comment in the paper, I think we should be fine:

Since there are no strong biological explanations yet to justify the causes of tritanopia
and tritanomaly, tritanomaly is simulated based on the shift paradigm only (Equation 4.19) as an approximation to the actual phenomenon and restrain the proposed model from try- ing to model tritanopia. Likewise, simulation of monochromacy (either rod or cone) is not supported by this model. One should note, however, that the conditions covered by the proposed model (i.e., anomalous trichromacy and dichromacy) correspond to approximately 99.96% of all CVD cases (RIGDEN, 1999; SHARPE et al., 1999).

I'll add a warning in that regard in our implementation.

What do you think?

[MichaelMauderer]

Looks good then!

[KelSolaar]

Thanks @MichaelMauderer!

2.5 years in the making :)