Yuhang Wu
Engineer ▏Designer
Arts & Computer Science
Color Transfer Software
Arts
Traditional Mural Arts in Shanxi & Museum of Mural Arts and Painted Sculptures
01
02
03
04
RESEARCH
Signal and Information Processing
Low-Bit Quantization for High-Resolution Wide-Swath DBF-SAR
Signal and Information Processing
Study on Target Imaging under Active Jamming Conditions for Wideband Phased Array Ragar
 Murals at Southern Part of Shanxi |
|---|
 Murals at Southern Part of Shanxi |
 Murals at Southern Part of Shanxi |
 Murals at Southern Part of Shanxi |
 Two Points ImagingAfter Range Compression and Amizuth Compression. |
|---|
 Results of BAQ Quatization |
 Results of 1 bit Quantization |
 Color Transfer DemoTop Left: Source. Top Right: Marked Source. Bottom Left: Target. Bottom Right: Marked Target. Right : Result |
|---|
 Differences in DetailsLeft : Target Right: Result |
 Demo 2Top Left: Source Bottom Left: Target Right: Result |
When simulating the imaging process of High-Resolution Wide-Swath SAR with the Range Doppler algorithm, the computational efficiency is low due to the huge data from multichannel received signals. The Block Adaptive Quantization algorithm, successfully implemented by the American Magellan SAR system, is much easier to be implemented for compressing the raw data that is from all radar channels to 8 bits.
I applied 1 bit quantization for both echo signals and range matched filters by deriving weights of harmonics, and then utilized the Lloyd-Max quantizer for 4 bits, 2 bits and 1 bit quantization. My evaluation software evaluated results of different quantization based upon the resolution, Integral Side Lobe Ratio and Peak Side Lobe Ratio. The results indicated that these methods reduced data rate and only slightly degraded image qualities.
Both projects of ”Traditional Mural Arts in Shanxi” and ” Museum of Mural
Arts and Painted Sculptures” are aiming at studying and preserving murals
that have historic and artistic value. Through digitization, 3D printing and
edits from masters of Chinese arts, we hope to present Chinese ancient arts,
especially mural museums, through remediation.
To assist experts in organizing and copying art works, we collected murals
from Huayan temple, which has almost 1000 years of history. In addition,
we analyze aesthetic judgments and artistic characteristics based upon
brushstrokes, colors and materials.
Instagram provides 10 filter effects for users to edit their pictures,
however, choices are limited. This project is an interactive tool to
transfer the color statistics from user-specified regions to target
regions. Thus, this tool is able to extract various filter effects from
different images. I developed this system in C++.
It utilizes a global color transfer method that is effective when two
images have similar color compositions and distributions. I employed
a color space conversion to convert RGB values to Lab values, which
maintain the wider color gamut. To process local regions, I utilized the
previous method on specified regions, and then applied the Poisson
Image Editing and the Gauss-Seidel SOR method for global
optimization. Both methods helped maintaining the gradient by
keeping salient features for visual quality, but resulted in unexpected
modification on other areas. In the next step, I plan to set boundary
conditions for global optimizations based upon marked unchanged
regions.