Digital Scanning: Testing the waters of AP 186

Back in the 20's, access to computers were very, very, very minimal, and as a consequence, scientists, engineers, chemists and all those people have to illustrate their graphs hand-drawn. Thus, its very tedious for the one making the graph to make the graph and also, people who are interested in analyzing the graph in detail have access only to the little amount of data included in the paper where in the graph is presented. Not to mention the errors that could surface from human technical errors (I'm not saying that there are errors, I'm saying that errors are a possibility).

With that in mind, and being that I am in an Image Processing class, Dr. Soriano assigned us to grab a copy of one of the many hand-drawn graphs of the 20's. I got the graph below from the Journal of Microbiology something something (Please excuse me for I have simply forgotten exactly where I got this graph, but don't worry, as soon as I recall where the graph has been uplifted, I will update this blog entry).

Figure 1: Hand written graph uplifted from a Journal of Microbiology dating back to the 1920's.

Now, What am I to do with this graph? Well, I have to numerically reconstruct this graph. How? Well, here comes the interesting part of the activity. By using the pixel information of this image, I was able to gather enough numerical data points so that I was able to reconstruct the graph. Using the relatively primitive but still effective software MS Paint, I was able to grab a hold of the pixel information of this image. The image is 2149 x 972 pixels in size. For the x axis, for every 0.05 unit, 477 pixels are used, and for the y axis, for every 0.1 unit, 186 pixels are used. The (0,0) mark on the graph is found at the (104, 813) pixel location. Now with all these pixel information, all I had to do next is use ratio and proportion to determine the numerical values of the graph (An example on how to determine the numerical value would be like this, lets say the pixel location is (106,812), from the (0,0) mark of the graph, the true pixel location of the data point would be (2,1), and then for the y axis, (0.1/186)*1 = 0.000538 and for the x axis, (0.05/477)*2 = 0.00021. And thus the numerical value is (0.00021, 0.000538)). By moving my mouse about the trajectory of the slope, I was able to gather enough numerical data points and thus figure 2 was born.

Figure 2: A quick comparison of the reconstructed graph with the original hand drawn graph

As you can probably observe, there is quite a small mismatch between the reconstructed graph with the original one. Well this is probably due to any distortion to the image that arose from the time the graph was photocopied then digitized using a scanner. Distortion causes the image to basically appear irregular, or in this case, I got the digitized graph a bit slanted and thus small anomalies followed, but from what I see, I would say its about 95% to 98% match, which in my book,is pretty awesome. So, for reconstructing this graph, I would probably give myself a grade of 10 out of 10, basically because I understood what I did, and I enjoyed doing it.

For the activity itself, in the lore of technical correctness, I'd probably give myself a grade of 5. And for quality of presentation, because of the unfortunate thing that is the distortion, I only deserve a maximum grade of 4. So, even though I couldn't give myself a perfect score, I still had fun with this activity. Rock on.