Bio

Qin-Qin Lü (pronounced Chin-chin Lyu) was born and raised in a river town in inland China. He holds a bachelor’s degree of physics from University of Science and Technology of China, and is currently a PhD candidate in physics at Louisiana State University. His research focuses on ultracold atoms, which are atoms that exhibit quantum behaviors such as Bose-Einstein condensation at extremely low temperature. He is a huge enthusiast of archaeological sites and ancient ruins. He loves to talk about technology advances and their role in history. He once bragged about a half-marathon, which he completed poorly. He loves food, but his stomach is most loyal to Chinese cuisines.

Project: Object Recognition in Aerial Images: a First Exploration

Identifying objects in aerial photos is important in many georelated fields, such as archaeology, agriculture investigation, urban design, and climate change study. Intuitively, people depend on geometrical shapes, terrain, morphology, color, and contrast to identify different objects. To process large amount of images in industry and research, computerized methods of object recognition are needed. In this project, I try to develop codes that identify objects by mimicking what humans do. After gamma correction and data format normalization, the images are analyzed by an array of filters, including shape filters (to detect lines, curves, corners, etc.), terrain filter, and color filter. Some of these filters apply to the entire image before or after morphological transformation; others apply to image segments, which are obtained from image segmentation by color. Based on the result of filters, images are then analyzed to determine whether the area is urban, suburban, or rural. For example, if many straight lines and rectangles are detected, and black pixels that are possibly shadows of buildings occupy a fair fraction of the image, then there is a high possibility that the image depicts an urban landscape. Automatically creating maps tagged with usage of each object with high accuracy and high efficiency is the ultimate goal in this field.

Favorite Four-Color, Four State Turing Machine

Rule 268913976372430922572919