Using ExoPlayer, developed a customized video player, which can have borders, can apply image processing filters to each frame using OpenGL fragment shader. Some comments: ExoPlayer uses GLES 2.0, in order to support OES_EGL_image_external.

However, GLES 2.0 only supports texture of 8 bit integer. For a look up table of image coordinate distortion, 8 bit is not sufficient. A workaround is to have 2 separate textures of 8 bit integer, then assemble them back to a floating point number of a higher accuracy.

Another small Android app, decodes and processed each frame of an input video file and then encodes frames back to a new output video file (using MediaCodec). It seems MediaCodec is not well documented. I used the code from media/cts/ExtractDecodeEditEncodeMuxTest.java

Learn about needs of customers.

Communicate with customers in an efficient way.

Use different sentences to express a similar meaning: for example, use passive tense; do not be judgemental even though some customers do not use the software in a correct way; more pleasant words.

Happy face, happy mood. Research indicates that if you put on smiles often, eventually your mood will get better.

Refer to: Ron Gutman: The hidden power of smiling | TED Talk

Prototype using Matlab, product using C++ and QT.

My work on 2D Image Processing Using OpenGL includes:

Implemented programs of color manipulation on color images, using all kinds of color spaces (It was a lot of fun to read books and articles on color theories and color manipulation. This is a very good example about how to choose coordinates to manifest a complicated system to serve different purposes). 

Edge directed interpolation; 2D DCT filter, 3D DCT filter for image denoising; the guided filter for image denoising, upsampling.

(BTW, I really enjoy viewing colorful colors :) But I also enjoy sketches only using black pencils.)

Developed an Android application for horizontal 360 degree panoramic videos 

Developed augmented reality (AR) horizontal 360 degree panoramic video for Android, using FFmpeg and OpenGL.

Of course, this is not a new technology. There are many products on the market.

You could take a look at these amazing 360 degree videos on the Internet. 

Our product may add some new flavours.

Developed check optical character recognition (OCR) for Android, using Tessearct OCR and OpenCV.

This application is based on open-source tessearct OCR. I trained the OCR, added a new API in the tessearct OCR jni, and applied an adaptive binarization algorithm to the photo of the document. It works very well.

Ying Li, Giovanni Gregori, Byron L. Lam, Philip J. Rosenfeld, Automatic montage of SD-OCT data sets, Optics Express, 19(27), 2011. (pdf)

Ying Li, Giovanni Gregori, Byron L. Lam, Philip Rosenfeld. Automatic algorithm for the Montage of Spectral Domain Optical Coherence Tomography (SD-OCT) Datasets. ARVO 2011, oral presentation. (2011)

Ying Li, et al. "Ridge-branch-based blood vessel detection algorithm for multimodal retinal images." Medical Imaging 2009: Image Processing. Vol. 7259. SPIE, 2009.