My first published scientific paper

research paper

Last month, my first scientific paper, Parallel Image Registration Implementations for GMTSAR Package , get published on the journal Seismological Research Letters. It is based on the work of my bachelor thesis at the School of Earth and Space Sciences, USTC.

In fact, I had confirmed to change my major to ECE before doing my bachelor thesis. But I still had to do it at my original department because I was going to get the degree in geophysics instead of ECE. I choosed to stick to topics related to my future major (image processing or whatever programming jobs). It happened that I was doing some work in InSAR image processing with two other schoolmates(my jobs are mainly helping them setting up softwares), so it would be convenient to focus on the same topic without digging into new geophysical topics (which are likely to be useless for me in the future).

I then talked to Prof. Zha Xianjie and he provided me a feasible topic, which is to convert a fraction of GMTSAR (an InSAR processing software) to parallel programs. Most parts of GMTSAR are serial programs so it takes no advantage of recent multi-core CPUs. I am quite satisfied with the topic, as it is pure programming job and I can learn some basics of multithread programming in C. Prof. Zha suggested me to focus on image registration module (mainly xcorr program), because it is a critical and time-consuming step in InSAR processing chain.

Finally, I finished a multithread version of xcorr program in C and a parallel image stitching scripts. These are all my work of bachelor thesis. I released the code on GitHub. Very simple parallelization, no much related to geophysics.

At the last thesis presentation (on June), I was the last one to present. Surprisingly, our Associate Dean, Prof. Yao Huajian was much impressed by my work. He encouraged me to publish an paper about it on some journal. So I began my first English paper under the supervison of Prof. Zha.

Although professors gave high comments on my work, I still doubted whether these simple programs are qualified to be advertised on a scientific journal. Without any instructions, I write another version of parallelized xcorr running on GPU with ArrayFire library (which may use CUDA or OpenCL as backend). I just want to enrich my work a little. (The code is on GitHub as well)

It takes about a month or so to finish the paper. Prof. Zha gave a lot of revision comments. Actually we spent most of (maybe 2/3) time on small refinements. The paper was finally submitted to Seismological Research Letter at the beginning of September.

In November, we received revision comments from SRL. These comments are not positive (in my opinion), but Prof. Zha said negative comments are just routine of paper review. So I edited the paper according to each comment, and he wrote the reply Email for me.

Finally, my paper got acceptance in late January this year, which was really incredible (before this I had no confidence at all). It was published last month on the website of SRL. Many thanks to Prof. Zha for his supervision.

Seemingly, this paper becomes my last link to geophysics. Hopefully my code does help anyone.

Page created on 2018-03-30