Motion-Compensated Noise Reduction of B &W Motion Picture FilmsBackground/MotivationCharacteristic of Blotchy NoiseProblems & ChallengesProposed SchemePre-processingJoint Motion/Noise DetectionMotion Trajectory EstimationPost-processingResult DemoSlide 11Slide 12Slide 13Motion-Compensated Noise Reduction of B &W Motion Picture FilmsEE392J Final ProjectZHU XiaoqingMarch, 2002Backgro und/Mo tiv ati on•Digitization of conventional video data• Achieving motion picture films•Major artifacts of B&W motion picture films:•Blotches: “dirty” spots and patches•Scratch lines•Intensity instability(illumination fluctuation) …•Previous work•General denoising: joint filtering•Line Scratch: model-based detection & removal •Blotchy noise: seldom addressed specificallyMy WorkCharacteristic of Blotchy Noise•They are: •Arbitrary shape & size•Obvious contrast against background•Non-persisting in position•They might NOT:•Be purely black/white•Have clear borderTypical BlotchesProblems & Challenges•Huge amount of data•Restrict computational complexity•Automatic processing preferred•Motion estimation tricked by :•Presence of noise•Illumination Change•Blurry scene for fast motion•…•Automatic detection not easy•Blotchy noise not readily modeled•Decision rely on motion compensated resultsProposed SchemeBlotch DetectionMotion DetectionMotionEstimationWrite out FramesRead in FramesMCFilteringTemporal Median FilterSection-wise Pixel-wise Frame-wiseWindow=5 ‘sandwiched’ABPre-processing•Five-tap temporal median filter•Effectiveness:•Generally denoising the sequence•Already removed blotchy noises•Introduced artifacts •Blurring of spatial details at regions w/ motion•missing fast moving linesJoint Motion/Noise Detection•Section-wise scanning of each frame•8*8 sections, non-overlapped•“sandwiched” decision-making•Two stage detection:•1st step: “change” detection•Criterion: Mean Absolute Difference(MAD) & “Edgy Area”•Original frame vs. filtered frame•2nd step: motion or noise•Criterion: ratio of MAD (should be consistent)•Reject changes due to blotchy noiseMotion Trajectory Estimation•Only computed for detected sections •Dense motion vector field estimation •Block-matching: •Neighboring block for each pixel: 9*9•Translational model •assuming smoothness of MVF •Full search• search range (-16, +16)•weighted MAE criterion•Error weighted by reciprocal of frame difference (A-B)•rejecting noisy dataPost-processing•Goal: remove artifact with MC-filtering•Available versions of the frame•Original•Temporally median-filtered•Motion compensated (bi-directional)•Modification strategy:•Linear combination•Median filter (spatial/temporal/joint)•Hybrid method (with edge information)Result DemoResult DemoResult DemoResult