Friday, July 29, 2016

Day 18

Today was a short day for me since I had to leave early for a trip over the weekend. I was able to work on my Gough Map chips in the morning. More PCA, SAM, and classification. So really, nothing new.

For the lunch cookout, we had kielbasa and Italian sausages, grilled to perfection by Emily. I was there to give her useful grilling tips and advice. (This is sarcasm, I only distracted her.) But in all seriousness, Emily did a great job on the grill, considering she had little previous experience.

The interns attempted another volleyball game. Before long, we succumbed to the heat and the fact that we struggle as volleyball players and decided to play frisbee in the shade instead. We decided that frisbee would be a more forgiving and less physically taxing activity.

Thursday, July 28, 2016

Field Trip Day!

Today, we finally got to go on the field trip we had planned a couple weeks ago. I was surprised by how nice our transportation was. The bus had comfortable seating, wifi that worked briefly, and strong air-conditioning.

The Eastman museum tour was about 2 hours long. We saw a room with old cameras and famous pictures (and had a great tour guide), then we saw the conservation room. The conservation tour guide seemed a tiny bit scared that we would touch everything with our grubby fingers and mess up the state-of-the-art restoration facility. He talked about the different processes involved in preservation, which was very interesting. I really wish we got to see him restoring a document, though.

Those galleries were the only two we got to see. Oh well...I was starving anyways. We got to go to Amiel's afterwards via the nice bus. The best part was realizing that they had chocolate milk there.

It was so hard to focus on working after we got back to RIT. I did make progress, though, and reached chip 28/32 by the end of the day. Not bad. If I'm extra efficient, I'll finish all 32 tomorrow.

Wednesday, July 27, 2016

Day 16: Blur, Divide and Conquer

This day was quite slow. I think back to this morning and it feels like it happened a year ago. I recall a discussion about Friday cookout meat options (Sausage? Shark?) and some frustration due to movie night planning issues.

I continued to process Gough Map images like a robot. I don't mean this in a negative way. Actually, I enjoyed the feeling of mechanical efficiency as I powered through image after image. 

Anna stopped by before lunch and answered some of my questions regarding Blur and Divide. We tested this method different ways to figure out its constraints. A lot of times, we got funky results. It was all part of the learning, though, and helped us get a better sense of how to use this method for optimal results. Thanks to Anna, I am now ready to blur, divide and conquer all of England. 

I was so occupied with figuring out Blur and Divide that I didn't have time to eat pizza before the lunch talk. I can't believe I put ENVI before food! This program must really be growing on me. 

The lunch talk was cool. I want to try the virtual game where you have to catch a ball with a paddle. Wait, no I don't. Because I would be the person who misses the ball 100% of the time.

Tuesday, July 26, 2016

Day 15

Contrary to popular belief, I was not at a doctor's appointment this morning.

I arrived around 10:00 and saw that there were donuts, but was disappointed to see that somebody had eaten the Boston Cream (whoever you are, you must confess). I officially finished splitting up the Gough Map into different chips and have a total of 32 now, many of which overlap.

For some reason, I find it necessary to list out what I must accomplished in the near future:
  • Work on the water sections of the map (which Allyse has mainly been doing but I want to get a taste of it myself)
  • Perform PCA, PCA with masking, SAM, ACE, and MF on all areas of interest (towns where original text is still visible)
  • Blur and divide these areas along with the results from the methods above 
The goal: separate the different inks. We're guessing that there are 3 types of red inks and we want to see which method(s) work best in separating them. There are only 3 bullets on my checklist, but I think this will keep me very occupied for a while (especially bullet 2). I think ENVI is growing on me though...and I have also mastered an efficient naming system for my hundreds of files created thus far.

I missed out on some Pokemon action during lunch...but got to chill with Maria instead, so not my loss.

Monday, July 25, 2016

Day 14: ENVI Crash Courses Cont.

Maria says that everyone is dead today. Maybe the thunderstorm contributed to that, or maybe it's the fact that it's a Monday.

Anyways, I have been bombarded with a bunch of new ENVI methods so let's see how well I can explain them.

MF: On a spectral graph, it measures how far a perpendicular line on the target pixel's vector must move to encompass another pixel so that it is considered similar to the target pixel. (Meaning that the pixel is on the same side of the line as the target.)

ACE: Like MF but slightly more accurate because it uses a cone in addition to the perpendicular line.

Did any of that make sense? Hopefully.

During lunch, Bob helped me open a cabinet (that didn't have a handle...) so that we could play a board game. We chose Trivial Pursuit. I was disappointed in myself for forgetting that the Hubble Space Telescope was a thing.

Later in the day, Allyse, Di, Dr. Messinger and I took a field trip to Anna's lab. It was a cool little hideout cluttered with old imaging instruments and hard-drives storing, like, a billion GB of ancient document data. Anna showed us how she was working on spacial data (while we are working on spectral.) She taught us an ENVI method that sharpens writing/structures on images. It has the coolest name yet: "Blur and Divide." It makes me think of "Divide and Conquer" even though they really have nothing to do with one another.

Friday, July 22, 2016

Day 13 (counting skills on point)

My previous assertion to finish analyzing the 15 Gough Map chips today was not fulfilled. Regardless, it was not a completely unproductive day, considering that I had spent the majority of the morning as a test subject for an RU's music experiment. John (the RU), Cici, Emily, Allyse, and I had trekked out into the oppressively warm outdoors to seek out a rare, piano-containing room far, far away. Once we reached our destination, we spent a few minutes (or maybe an hour) tuning the violins and picking out the best violin/player combo for the experiment. John recorded us playing different notes on the violin and piano simultaneously. Later, he will attempt to separate the instruments in the recording and determine the notes played by each using some complicated method of his. This experiment also doubled as a start to our internship orchestra. Next stop: Carnegie Hall.

The lunch cookout was quite a messy experience. First, we found ourselves lacking a proper watermelon-cutting knife and had to resort to plastic knives and Cici's Swiss knife. Barely hindered by our lack of resources, I skillfully cut the watermelon with the Swiss knife and have decided to pursue a career in the food industry (McDonald's?) Meanwhile, the watermelon soaked through about 80 thin paper plates. The lunch mess continued with a frenzied scenario in which a giant bug dropped into the ketchup on my plate and only moments later, a gust of wind blew the plate of ketchup right into Cici and me, covering our arms in ketchup. The mess was unbearable, and maybe kind of funny, but we figured it was a good enough excuse to skip volleyball practice for the day.

And now I will talk about actual important things, like the Gough Map. I got to chip 14 out of 15, and I am making an achievable goal for myself to truly finish all 15 chips on Monday. I also look forward to separating the different inks (original/revised) on the entire map, which is our next project, along with the Mexican codex.

At the end of the day, we began to speculate on each of the intern's future. I will be interested to see which of our predictions come true.

Thursday, July 21, 2016

Day 12

This interesting day began when Bob launched us into a discussion of "the worst thing that happened during the school year" at each of our respective schools. We all went into depth about senior pranks and other horrors of high school, then departed from the morning meeting to go do serious stuff.

This serious stuff, for me at least, included getting a NEW DOCUMENT TO WORK WITH! It was a Mexican codex that had been showered in chalk-like paint at some point in its approximately 5 century life. What remained visible was...pretty much nothing. Dr. Messinger and Di warned Allyse and me of the difficulty of extracting the underlying structures in this document. Although it was a daunting task to undertake, I was up for it. I mentally prepared as I waited 30 minutes for the image to download on my computer.

This may be a disappointment, but I did not end up working on the Mexican codex today. Allyse's sensible voice guided me into going back to the Gough Map documents and performing more thorough analysis on them before moving on. (No offense, Gough Map, you're really cool but I just wanted to try something new.) I owe Allyse for this insightful advice, though, because I made much more progress on the map in areas I had overlooked previously.

With an unnecessary amount of difficulty, the navigationally challenged interns made it to the Student Union for lunch. Nothing too funny happened there.

For the rest of the day, I sat at my usual computer in the Reading Room with a fierce determination to perform PCA and SAM on all 15 chips I had created of the Southeast section of the Gough Map. Sadly, I only made it to chip 11 (second disappointment of the day), but I believe I produced the most results today out of all the days so far (so that cancels out one of the disappointments!) I plan on finishing this task tomorrow.

Wednesday, July 20, 2016

Day 11

Today, we organized what food we would be bringing to the Friday cookout. After the morning meeting, Allyse and I checked in with Dr. Messinger and continued to work on the southeast section of the Gough Map. Allyse found some interesting letters and pictures in the sea surrounding England, which was cool. I started putting my results into a powerpoint for organization and documentation. I also helped Cici look for paris green, but was unsuccessful.

We got to enjoy free pizza for lunch and also attending an interesting talk on machine learning and recognition. I realized that this area of study is very appealing and promising (people being offered $1 million starting salary right out of college!!!) Maybe I should look into this as a future career...

Tuesday, July 19, 2016

Day 10

Today consisted of more work on ENVI. Allyse and I had picked the Southeast section of England (the area with most revisions) to focus on. I made 6 more chips on the land part of the map while Allyse worked mainly on the water-surrounding-England part (we are both lacking in geography skills so we're not sure what body of water it is...the Atlantic?)

I pulled out writing from several of the chips and also found differences in the ink used for the rivers. We are unsure of the significance of this, but slight changes in ink may indicate tides or a certain water depth. 

For example, there are differences in darkness along the rivers in this image. The dark spots could be intentional or they could just be areas where the ink was drawn on thicker.

To get this image, I selected one green pixel from the water and used SAM to show which pixels in the image are most like that selected green pixel. The darker areas represent pixels that are more similar to the green pixel. Clearly, there are differences in the color/thickness of the ink along the rivers.

At 4, we attended an entrepreneur talk about speaking to non-science audiences. We learned how to be good storytellers and integrate a storyline into a scientific presentation in order to make people care about whatever we are sharing.

Monday, July 18, 2016

Day 9: Statistics

In the morning, we had a peer review of our abstracts where we gave and received constructive criticism. Using the suggestions given to me, I later revised my abstract.

After the peer review, Allyse and I met with Dr. Messinger and learned a new ENVI technique called "RX Anomaly Detection." This method uses statistics (standard deviation) to identify any anomalous data (pixels) in the image. It represents the level of difference of each pixel from the norm in a grayscale image. In this image, the lighter areas are considered closer to "anomalies" than the darker areas.

Allyse and I were also given a new section of the Gough Map and pretty much set free to work on it. We learned how to make our own chips (smaller sections of the map) and were allowed to use any computer processes to enhance them. I made 8 different map chips and experimented with RX Anomaly Detection and also incorporated some SAM and PCA.

In the afternoon, I participated in the Visual Perceptions Lab experiment, but I can't say anything else about it because the experiment is top secret.

Friday, July 15, 2016

Day 8.5: Abstract

Madeline Loui and Allyse Toporek

Enhancement of Hyperspectral Imagery of Historical Documents

Several weeks ago, Dr. Messinger attended a conference at Oxford University to present new computer processing methods to enhance the faded and illegible writing on images collected of historical documents. The work he presented was of images the university sent to him of the Gough Map of England. This document is around 600 years old and has undergone much work (updates) since its first creation. However, some of the ink has faded with time or other environmental factors and what remains of that ancient writing and pictures are grey, organic, unrecognizable structures or nothing at all.

We have worked to learn those computer processes and programs in the ENVI software and custom processing code created by Di Bai, a PhD student in CIS. We are using them on smaller sections of the overall Gough Map in order to identify the original, hidden texts and figures. These include, but are not limited to, Classification processes, Principal Components Analysis (PCA), and Spectral Angle Mapper (SAM). Our finished works will be sent along with those of Dr. Messinger and those of Di Bai to Oxford for analysis and hopefully, description. A broader goal is to assess the different methods and their capabilities.

Day 8: A Pretty Fly Day

This morning, Allyse and I worked on our 89 slide powerpoint in which we included all of our best ENVI results so far. Dr. Messinger will be sending the powerpoint to Oxford in the near future and then we will wait for feedback.

Cici and I showed off our volleyball moves during the cookout lunch. We had just come back from the Olympic Trials, so we were in tip-top shape. The other interns were BLOWN AWAY by our skill.

While training for the Olympics, we also started a state of the arts paper airplane factory, with the help of Niels. Our handcrafted, one-of-a-kind paper airplanes reach velocities up to 1 m/s and cruising altitudes of about 5 feet! The latest version also twirls through the air (this feature can be bought with an additional $8000 fee.) Cici and I have thoroughly tested our airplanes by throwing them from 4 stories high; the airplanes remained in fabulous shape and it is safe to say that their quality is guaranteed.
Our original models (starting price $50000)

In the afternoon, Allyse and I made a few last minute additions to our powerpoint. Our slide count rose to 99 and frustrated by the number, I added a slide that said "Thank you" just so that our powerpoint could be 100 slides long. With the PowerPoint passing our approval, Allyse and I gave the file to Dr. Messinger to give to Oxford.

Thursday, July 14, 2016

Day 7: It's Almost Friday Yesss

I have 2 shoutouts today: one to Mr. Callens for bringing us donuts and another to Maria who asked what an abstract was, because deep down, I wasn't quite sure either but assumed that everyone else knew so I avoided asking the question myself.

Anyways, more work on ENVI today. Mostly redoing the images from yesterday. This time SAM worked and didn't cut off the images, so that was a plus. I organized the over 50 slide long PowerPoint of our results that Allyse and I started yesterday.

So since SAM worked today, I can show some results.

Original:                                                       Processed:

Easier to read? I think so!

For lunch, all the interns walked to Global Village (after attempting to decipher a map and asking a pedestrian for directions.) I appreciated the change in scenery and my burrito bowl.

Afterwards, Allyse and I went back to our computers, whipped up a rough draft for our abstract, and did other techy computer stuff. I finished my part on ENVI for the day early (SAM on images 1-5), so now I am blogging...and doing random tasks (filling up Emily's waterbottle)...and apparently stealing cookies. Allyse, Emily, and I also took a walk around the building and found all the other interns. Cici and Zihoa gave us a demo of their lab and we got to look at their colorful samples.

Wednesday, July 13, 2016

Day 6: SAM the Cat

This morning, Allyse and I learned a new method to use on ENVI. It's called SAM, which stands for Spectral Angle Mapper. Also, the guy who came up with it named it SAM because his cat's name was Sam! SAM can help make differences in ink color more obvious by comparing one target pixel (that is selected by the user) to every single other pixel on the image. Then, it represents the level of similarity of the pixels in a gray-scale image (with black being closest to the target pixel and white being furthest.)

After learning this method, Allyse and I applied it to the Gough Map images. Sadly, the resulting images were cut off and we were unsure of how to fix it. Still, we got a some considerable results that brought out the faded writing.

We attended the lunch talk about visual perception and the concepts behind eye movement. I enjoyed the interactive parts of the presentation and learning that the eye rotates 3 different ways (not just up/down and left/right!)

After lunch, Allyse and I started piecing together a hodge-podge powerpoint of our accumulative Gough Map results since Day 1. We will continue to organize the powerpoint until it is presentable to the historians at Oxford who are looking forward to seeing the results!

Tuesday, July 12, 2016

Day 5: Prefominanylu

To begin, it is important to announce that Cici and I have selected this lucky creature to be our internship mascot. His name is Prefominanylu. He keeps us company at the computer.

Anyways, today was rather productive. Di stopped by the Reading Room to give us a quick lesson on how coding helps make PCA more effective. He had written a code on Matlab that enabled insignificant parts of images to be blocked out so that PCA could focus on the significant parts. We didn't learn how to write any code (not yet), but we learned how to apply Di's code onto the images of the Gough Map and use the skills we had learned yesterday (PCA) to further manipulate the images.

I ran the code on 9 different sections of the map and enhanced the resulting images so the hidden writing became more prominent. Hopefully, some of my results will be useful to the professors/historians at Oxford!

Allyse and I were on the computers for what seemed like forever, but we actually had several nice breaks. First, the surprise fire drill happened. Second, we experienced our typical (and predicted) mid-afternoon crash and felt the urge to go outside and lie down on the grass. So that's what we did. I also read aloud a couple pages from Lord of the Flies to Allyse as we were getting fried by the sun. (Hopefully it wasn't dreadfully boring.)

As you can see, my day was pretty prefominanylu! (@ Cici)

Monday, July 11, 2016

Day 4 (Our Very First Monday)

We started off the day by saying a bit about each of our labs to the other interns. I was very interested in what the others were working on.

Afterwards, Dr. Messinger taught Allyse and me how to use PCA (Principal Components Analysis) on ENVI. Pretty much, PCA detects a coordinate system that better fits the data (pixels) of the Gough Map images. Then, the image can be represented in a different way and reveal hidden text/structures. We practiced using PCA on the Gough Map images for a long time.

Allyse and I had been staring at the computer screen for many consecutive hours and had to take a couple breaks in which we walked around the building (in which Allyse complained about the annoying sound my shoes made when I walked, haha.) By the end of the day, we had many results from using PCA and had gotten the hang of the process. Soon, we will be learning how to improve results by writing code to make PCA more effective...yay!

Friday, July 8, 2016

Our Very First Friday

This Friday, I had much progress on ENVI and learned multiple new classification methods (thanks to Di.) One of the images I classified using Maximum Likelihood revealed structures and writing on the Gough Map that were not clearly visible in the original map. Messinger informed us that we were the only people using this method to analyze the Gough Map and that our results were of great significance to the owners and historians dealing with the map over in England. Although the image will need to be processed further (in Photoshop) in order for the writing/picture to be deciphered, the classification methods have shown considerable progress in the process.

Here is the Gough Map image after Maximum Likelihood was applied with the areas of interest circled.

Here is the original image; it is very hard to make out the structures that appear in the edited image.

Also from today was our first cookout lunch, which was enjoyable despite the rain. I contributed by putting the cheese on the burgers. We attempted to have a fast-paced, competitive volleyball game. Thankfully, Cecelia and I (both with years of experience on varsity teams) were there to keep the ball in the air. We practically saved the group from looking like unathletic nerds.

Thursday, July 7, 2016

Day 2: Starting ENVI

Allyse started off the day by organizing the food sign-ups for the Friday lunch which was really proactive of her. Afterwards, we met with Dr. Messinger who told us to experiment with ENVI so that we could get familiar with the program. He also told us about ENVI's origins, which was created in a garage by 2 guys in Colorado.

We decided to attend one of Dr. Messinger's student's thesis defense on an innovation about remote sensing. The most difficult part was getting into the room; we arrived late and the door was locked so we were watching the presentation through a window, which was probably really creepy. Thankfully, Joe Pow saw us spying on the presentation and unlocked the door for us so we wouldn't have to be creepy. During the defense, I had trouble following what the PhD student was talking about, but I could tell she was very knowledgeable of her topic. Overall, it was interesting to see what a defense looks like.

Afterwards, Allyse and I got started on ENVI. I classified the 1st image of the Gough Map that Di gave us 9 different ways and documented them in a powerpoint. Here is an image of the Gough Map after I applied unsupervised (K-Means) classification, which instructed ENVI to separate the image into 10 categories of color.

At the end of the day, Allyse, Di, and I met with Dr. Messinger and looked over some of the results. We will continue to experiment with the software for the next few days.

Wednesday, July 6, 2016

First Day!

To start, we got a tour of the imaging center, given by Joe Pow. Then, we walked to the Red Barn for team building games. The barn must've heated up from our intense brain power because it was a million degrees in there. The games were fun and challenging and I got to know the other interns better. When we finished, we walked to lunch in the boiling heat.

Afterwards, we met with our advisors. Allyse and I talked to our advisor, David Messinger, and met the grad student we would be working with, Di. Dr. Messinger told us that the imaging techniques he presented to scientists at Oxford were more appealing to them than expected. They want RIT to analyze 2 documents, the Gough Map (of England) and the Archimedes Palimpsest. Di showed us some basics of manipulating the document images in order to reveal hidden writing/pictures in them. We experimented with the program, ENVI, and practiced color classification and linking.

It was a great first day and I'm very excited about this internship!!