What are Good Apertures for Defocus Deblurring? Changyin Zhou Shree K. Nayar Columbia University ICCP 2009, San Francisco.
Download ReportTranscript What are Good Apertures for Defocus Deblurring? Changyin Zhou Shree K. Nayar Columbia University ICCP 2009, San Francisco.
What are Good Apertures for Defocus Deblurring? Changyin Zhou Shree K. Nayar Columbia University ICCP 2009, San Francisco Geometry of Defocus Lens Object Aperture Pattern Sensor PSF Focus Plane Formulation of Defocus In the Spatial Domain Focused Image PSF Image Noise Captured Image Formulation of Defocus In the Fourier Domain Focused Image PSF Image Noise Captured Image Various Aperture Coding Techniques Annular Pattern [Welford 1960] [Levin et al. 2007] Wave-front Coding [Dowski 1993] MURA Pattern [Gottesman 1989] [Veeraraghavan et al. 2007] What are Good Apertures for Defocus Deblurring? Outline - Background - Evaluation Criterion for Aperture Patterns - Pattern Optimization - Experiments How to Evaluate Aperture Patterns? Natural image Gaussian white noise Captured Image Focused Image An Optimal Deblurring Algorithm Deblurred Image Basic idea: Evaluate K using expected quality of deblurred image How to Evaluate Aperture Patterns? Natural image Gaussian white noise Captured Image Focused Image An Optimal Deblurring Algorithm Deblurred Image Linear system The Optimal Linear Deblurring Algorithm For white noise and L2 distance, Weiner filter is optimal Optimize C by minimizing the expected recovery error The Optimal Linear Deblurring Algorithm log(A) : Gaussian noise level : Expected power spectrum of natural images A variant of Weiner deconvolution 1/f law How to Evaluate Aperture Patterns? Gaussian white noise Natural image Captured Image Focused Image An Optimal Deblurring Algorithm Deblurred Image Linear system Evaluation Function: Evaluation Criterion for Aperture Pattern Noise Level Frequency R(K | Prior Power Spectrum Expected deblurring quality at frequency ) : Expected deblurring quality at noise level . Evaluate Patterns Using the Criterion R Curve of Circular Pattern R 2 R( K | ) 2 2 | K | / A Noise Level Evaluate Patterns Using the Criterion Curves of Relative R: R(K)/R(K0) R( K ) / R( K0 ) Veeraraghavan Levin Noise Level Annular Image(binary) Image(gray) Random • Coded apertures help more when noise level is low. • Circular aperture is better when noise level is high. Outline - Background - Evaluation Criterion for Aperture Patterns - Pattern Optimization - Experiments Pattern Optimization Difficult to solve analytically Patterns evaluated in the Fourier domain, but strictly constrained in the spatial domain . Difficult to do brute force search For binary patterns of resolution N x N, the number of possible solutions is huge, when N = 13, if evaluating one pattern takes 1 millisecond, the brute force search requires 1045 yrs. Pattern Optimization Pattern Evolution in Genetic Algorithm: 13 x 13 binary patterns; 8 different noise levels σ =0.0001 σ =0.001 σ =0.002 σ =0.005 σ =0.008 σ =0.01 σ =0.02 σ =0.03 2nd Run: 3nd Run: Evaluate the Optimized Patterns Evaluate the Optimized Patterns Relative R curves of the Optimized Patterns R( K ) / R( K0 ) Circular Noise Level Image Veeraraghavan σ= 0.001 Outline - Background - Evaluation Criterion for Aperture Patterns - Pattern Optimization - Experiments Implementation Precision Laser Photoplot (1 micron) Implementation Canon EF 50mm f/1.8 Lens Implementation Image Pattern Veeraraghavan et al.’s Pattern Levin et al.’s Pattern Circular Pattern (wide open) Our Optimized Pattern Comparison Experiments on a CZP Chart Captured Images Focused Image Circular Pattern Levin et al.’s Pattern Veeraraghavan et al’s Pattern Image Pattern Our Optimized Pattern Comparison Experiments on a CZP Chart Deblurred Images Focused Image Circular Pattern Levin et al.’s Pattern Veeraraghavan et al’s Pattern Image Pattern Our Optimized Pattern Shrek Captured Image using the Optimized Pattern Shrek Captured Image using the Optimized Pattern Deblurring Result On the Street Captured Image Using the Optimized Pattern On the Street Captured Image Using the Optimized Pattern Deblurring Result Traffic Scene Captured Image Using the Optimized Pattern Traffic Scene Captured Image Using the Optimized Pattern Deblurring Result Summary Main Contributions Aperture Evaluation Criterion for Defocus Deblurring Aperture Pattern Optimization for Defocus Deblurring Future Work Optimize gray-level patterns Apply Criterion to other PSF Engineering Problems Extend Analysis to Account for Diffraction Thank You!