Thomas Payne Jordan Key

Slides:



Advertisements
Similar presentations
Applications of one-class classification
Advertisements

Interactive Evolutionary Computation Review of Applications Praminda Caleb-Solly Intelligent Computer Systems Centre University of the West of England.
CSCE 643 Computer Vision: Template Matching, Image Pyramids and Denoising Jinxiang Chai.
CSC321: Introduction to Neural Networks and Machine Learning Lecture 24: Non-linear Support Vector Machines Geoffrey Hinton.
QR Code Recognition Based On Image Processing
Road-Sign Detection and Recognition Based on Support Vector Machines Saturnino, Sergio et al. Yunjia Man ECG 782 Dr. Brendan.
Computational Biology, Part 23 Biological Imaging II Robert F. Murphy Copyright  1996, 1999, All rights reserved.
Kriging.
Solar Feature Catalogues S Zharkov, V V Zharkova, S S Ipson, A.Benkhalil, N.Fuller, J. Aboudarham.
1 Video Processing Lecture on the image part (8+9) Automatic Perception Volker Krüger Aalborg Media Lab Aalborg University Copenhagen
Automatic measurement of pores and porosity in pork ham and their correlations with processing time, water content and texture JAVIER MERÁS FERNÁNDEZ MSc.
A Novel Scheme for Video Similarity Detection Chu-Hong Hoi, Steven March 5, 2003.
Detecting Image Region Duplication Using SIFT Features March 16, ICASSP 2010 Dallas, TX Xunyu Pan and Siwei Lyu Computer Science Department University.
Shape Modeling International 2007 – University of Utah, School of Computing Robust Smooth Feature Extraction from Point Clouds Joel Daniels ¹ Linh Ha ¹.
Xinqiao LiuRate constrained conditional replenishment1 Rate-Constrained Conditional Replenishment with Adaptive Change Detection Xinqiao Liu December 8,
Interactive Optimization by Genetic Algorithms Cases: Lighting Patterns and Image Enhancement Janne Koljonen Electrical Engineering and Automation, University.
Chapter 2 Dimensionality Reduction. Linear Methods
ENDA MOLLOY, ELECTRONIC ENG. FINAL PRESENTATION, 31/03/09. Automated Image Analysis Techniques for Screening of Mammography Images.
Multimodal Interaction Dr. Mike Spann
ALIGNMENT OF 3D ARTICULATE SHAPES. Articulated registration Input: Two or more 3d point clouds (possibly with connectivity information) of an articulated.
BARCODE IDENTIFICATION BY USING WAVELET BASED ENERGY Soundararajan Ezekiel, Gary Greenwood, David Pazzaglia Computer Science Department Indiana University.
1 Signal Processing Laboratory Swiss Federal Institute of Technology, Lausanne Cédric Dufour ( LTS-IBCM Collaboration ) The ‘microtubules’ project.
Compression of Real-Time Cardiac MRI Video Sequences EE 368B Final Project December 8, 2000 Neal K. Bangerter and Julie C. Sabataitis.
Measures of variability: understanding the complexity of natural phenomena.
3.7 Adaptive filtering Joonas Vanninen Antonio Palomino Alarcos.
GENDER AND AGE RECOGNITION FOR VIDEO ANALYTICS SOLUTION PRESENTED BY: SUBHASH REDDY JOLAPURAM.
By: David Gelbendorf, Hila Ben-Moshe Supervisor : Alon Zvirin
Team Members Ming-Chun Chang Lungisa Matshoba Steven Preston Supervisors Dr James Gain Dr Patrick Marais.
Digital Image Processing
A Framework for a Fully Automatic Karyotyping System E. Poletti, E. Grisan, A. Ruggeri Department of Information Engineering, University of Padova, Italy.
Logistic Regression & Elastic Net
SUREILLANCE IN THE DEPARTMENT THROUGH IMAGE PROCESSING F.Y.P. PRESENTATION BY AHMAD IJAZ & UFUK INCE SUPERVISOR: ASSOC. PROF. ERHAN INCE.
SUMMERY 1. VOLUMETRIC FEATURES FOR EVENT DETECTION IN VIDEO correlate spatio-temporal shapes to video clips that have been automatically segmented we.
Similarity Measurement and Detection of Video Sequences Chu-Hong HOI Supervisor: Prof. Michael R. LYU Marker: Prof. Yiu Sang MOON 25 April, 2003 Dept.
Computer Vision. Overview of the field  Image / Video => Data  Compare to graphics (the reverse)  Sample applications  Video Camera feed => ID room.
Reading and Writing Image Files
Machine Learning for Computer Security
Adaptive Median Filter
Hiba Tariq School of Engineering
Hand Geometry Recognition
Background on Classification
PRINCIPAL COMPONENT ANALYSIS (PCA)
Digital Image Processing Lecture 16: Segmentation: Detection of Discontinuities Prof. Charlene Tsai.
Digital Image Processing Lecture 10: Image Restoration
Supervised Time Series Pattern Discovery through Local Importance
IMAGE PROCESSING INTENSITY TRANSFORMATION AND SPATIAL FILTERING
Digital image self-adaptive acquisition in medical x-ray imaging
Elementary Statistics
Digital Processing Techniques for Transmission Electron Microscope Images of Combustion-generated Soot Bing Hu and Jiangang Lu Department of Civil and.
Anne Pratoomtong ECE734, Spring2002
Other Algorithms Follow Up
A New Approach to Track Multiple Vehicles With the Combination of Robust Detection and Two Classifiers Weidong Min , Mengdan Fan, Xiaoguang Guo, and Qing.
Detecting Artifacts and Textures in Wavelet Coded Images
IMAGE BASED VISUAL SERVOING
The Basics of Microarray Image Processing
Basics of Photometry.
An Introduction to Supervised Learning
10701 / Machine Learning Today: - Cross validation,
Lecture 3: Main Memory.
Midterm Exam Closed book, notes, computer Format:
On Convolutional Neural Network
Motion-Based Analysis of Spatial Patterns by the Human Visual System
Midterm Exam Closed book, notes, computer Similar to test 1 in format:
Presentation by: Lillian Lau
EE 492 ENGINEERING PROJECT
Introduction to Matlab
Midterm Exam Closed book, notes, computer Similar to test 1 in format:
A Novel Smoke Detection Method Using Support Vector Machine
Fracture Surface Analysis of Dual-Phase Steel
Review and Importance CS 111.
Presentation transcript:

Thomas Payne Jordan Key Extracting morphological changes in nanocrystals using in situ liquid cell microscopy Thomas Payne Jordan Key

outline Introduction and Motivation Type of Linkage Image Segmentation and Binarization Two-Point Statistics PCA Model Building Future Work

Introduction/Motivation Palladium nanocrystals from H2PdCl4 solution Nanocrystal size and morphology control properties, so being able to tailor the structure is highly desireable Liquid cell microscopy opens up a wide range of in situ experiments, such as corrosion and nanocrystal synthesis from solution, allowing for nanoscale resolution of real-time events

Process-structure linkage We aim to connect the process variables Electron beam dosage Exposure time To the structure variables Structural evolution of the nanocrystals

Data sets STEM videos of nanocrystal nucleation and growth BF and HAADF 3 videos: on the order of 10000 frames Input Electron beam dosage Exposure time Output (images) Microstructural evolution

Sample data frame Not all frames will be completely refreshed (beam raster) Some shadowing or blurring can be seen in the contrast

Frame extraction Developed an algorithm to make manual extraction easier Reduces number of frames to be manually searched from thousands to hundreds

Segmentation challenge Binarize gray-scale images containing two phases (liquid and solid) We explored several approaches: Simple global thresholding, adaptive mean thresholding, adaptive Gaussian thresholding, etc. For us, simple global thresholding was a poor solution

Segmentation solution – big disk filter Extract fully refreshed frames Currently manual but hope to automate Convert to grayscale → rgb2gray() Crop image Big Disk Filter → strel(‘disk’,25,0) Find average at every point in disk → imfilter(...) Remove background Remove noise via  median filter → medfilt2(...) Find threshold → imageSegmenter() Save binarized image → imwrite(im2bw(image,threshold)) Automatically process all the images going back and manually adjusting threshold as necessary*

Visualizing the big disk filter Locate and remove disk shaped “structuring elements” representing high signal regions (larger length scale than individual particles )

Sample binarized image

Automated python pipeline We now have routines that will load in all of the complete frames, segment and binarize them, and calculate two-point statistics *Used Otsu’s method to calculate the global threshold for each image, rather than manually adjusting threshold for each image Unfortunately we still have to manually extract the completely refreshed frames from the original data set, but automating that may be beyond the scope of this project

Two-point statistics Only need one auto correlation or the cross correlation Two-phase microstructure Non-periodic boundary conditions Visual inspection Volume fraction Radial symmetry

PCA For a long time ran into memory issues with the calculations Looked into using the kernel trick or some other method to address this, for a time just used PACE This issue has now been resolved and we can run PCA locally Utilized python module sklearn to compute PCA

PCA From the scree plot, we decide only first three pc scores are necessary, as the first pc score accounts for over 95% of the variance PC Score Percent Variance 1 96 2 2.5 3 0.5 Sum 99

PCA We now have three basis vectors and the corresponding pc scores for every microstructure in the array The first-term reconstruction captures most of the features of the two-point statistics, and the reconstructions get better with additional terms

PCA Our scree plot indicated that three pc scores should be sufficient for reconstruction, and this is justified by the three-term reconstruction below Comparing to the full two-point statistics, the reconstruction matches quite well

Model building We use a polynomial regression to fit our pc scores Using various metrics like R-squared fit and mean absolute error, we determined that a 5th order polynomial was the appropriate choice for the first two scores

PC1 PC2 PC3

Model building PC1 PC2 PC3

Potential Future work Improve upon the pipeline for automated analysis Reduce information loss from segmentation Rewrite Matlab code in python Use model to predict nanocrystal growth with time for new dosage and compare morphologies reconstructed from 2-point statistics to experimental results Connect morphology evolution with kinetic model from literature

Acknowledgements Questions? We would like to thank Dr. Ray Unocic from Oak Ridge National Lab for providing the data utilized in this project. Questions?

PC1 PC2

Matin suggestions User messages – currently no way to reply, no threads If I want to reply to a message, I have to go to someone’s profile and create a new message Automated messages from MATIN – no content With new forums posts being made, we are getting automated messages to our MATIN account but they have no content. A preview of their post or something would be nice Suggest automated messages for wiki comments – using above improvement If we got a message every time someone commented on our wikis like we do forums, that would be nice (if the message has content)