BioLumos
Living beings can interact with their environment. Some are sensible to different stimuli : for example, human beings can perceive light with their eyes. We, humans, also build tools to detect and measure different stimuli, for example light sensors (like cameras). During a week, we compared the characteristics of the human eye and of a RGB sensor, and we were trying to see how well are they able to perceive the difference between two colors.
In the human eye, the perception of different colors is possible thanks to the cones located in the retina. There is three kind of cones and each one of them is sensitive to one of these three fundamental colors : red, green and blue, which are enough for the human brain to have the sensation of all the colors. You can learn more about the human eye perception of colors through this very well explained TedEd video.
A RGB sensor is an electronic sensor that can distinguish colors. It contain photoreceptors for the same three fundamental colors (red, green and blue) and can make an average of those three colors depending on the environment.
In order to see how humans could perceive the difference between two colors, we took four slightly different oranges, similar but not identical, differing in the proportion of green pixels components. We randomly formed pairs with these oranges : The two oranges could be identical, slightly different, quite different or very different. The subject was is introduced in a dark room and had to sit on a chair facing a computer screen that displayed the following pattern of screens : 1s, 2s of black screen, 1s of second orange. This pattern was followed by 6s of black screen during which the subject was supposed to say if he perceived the two oranges as “same” or “different”. (Figure 1)
Figure 1 : The pattern of the video showed to human subjects
Believe me, the experiment was quite hard for the subject. We also showed the RGB sensor those four oranges to see if it could distinguish them. The whole documentation describing the experiments thoroughly, along with the data and scripts, can be found on Github.
The human subjects gave 66% of good answers, which is not a lot if you consider that 50% of these answer corresponds to “subjects answered randomly”. However, people contradicted themselves a lot : when facing twice the same pairs, they often gave opposite answers, like “same” the first time and “different” the second. On the total of 800 answers we had, 357 were contradictions ! (Figure 2). The RGB sensor detected the colors but didn’t noticed the difference between them.
Figure 2 : A global view on the answers given by human subjects
So, what are the differences and the similarities between electronic and biological sensors ? They are both sensible to red, green and blue. However, RGB sensors are very sensible to the distance to the object. Eyes are able to see pretty much the same within a few meters radius, while the sensor work very bad further away than 50 centimeters or so. Furthermore, the human eyes is not digital, meaning it does not give a numerical value associated with a color. RGB sensors are calibrated, they are able to associate a color with a number, which humans can not do. This is why we used comparison for humans (“Are those oranges different ?”) rather than asking a description, which is very hard for us. By the way, you could try to experiment on this : try to describe a color, only with human-readable words (no digits or diagrams) and tweet it to @bio_lumos. We’ll share the best ones !
Moreover, to be able to distinguish color isn’t only due to biological perception but also from social factors. For example, a researcher realized that color perception was also depending on your own language. See more in the following article : “Özgen, Emre. "Language, learning
© Lina Vigneron, Armelle Supervie, Samuel Churlaud
No comments:
Post a Comment