1Introduction 1Digital Image ProcessingECE 178B. S. MANJUNATHRM 3157 ENGR ITel:[email protected]://vision.ece.ucsb.edu/ManjunathIntroduction 2On the WEBFor course information:http://www.ece.ucsb.edu/~manj/ece178Teaching AssistantsPradeep KoulgiPratim Ghosh2Introduction 3Today Course outline Introduction to image processing Matlab basics and the image processingtoolboxIntroduction 4About this course Prerequisites– Strong motivation, basic calculus– MATLAB is the programming environment, but no prior backgroundin MATLAB is assumed. Who can take this course?– Juniors/Seniors/Graduate students in ECE/CE/CS/ME/MATP/… Reference Book:– Gonzalez and Woods, 2nd Edition (2002) or the new edition (2007)– http://www.imageprocessingplace.com– Read chapters 1 and 2 (available online)3Introduction 5Grading H/W* due by 5pm on due date Quiz (surprise!)* in class/discussion sessions– HW+Quiz 30% Midterms 20% (two mid-terms) Finals 50%*All homeworks/class participation are required.A non-submission will affect your grade non-linearly.You will get a fail grade if you miss morethan one h/w or prog. assignment.Introduction 6Why Image Processing? The future is multimedia informationprocessing…… Images (and video) are everywhere! Many and diverse applications– Astronomy, biology, geology, geography,medicine, law enforcement, defense, Industrialinspection,…– Different imaging modalities: visual, X-ray, ultra-sound, …4Introduction 7Entertainment Digital camcorders HDTV DVDs: High quality image/video compression(MPEG-2: about 5-10 Million bits/second) Digital Cinema– New compression technologies are needed– Consider a 2 hour movie: 1920 x 1080 x 30bits/pixel x 24 frames/second ~~ 1.5 billionbits/second  1.3 terra bytes / 2 hr programIntroduction 8Security Person Identification– Face recognition– Finger print identification Watermarking– Copyright protection and authentication Data hiding– Secret communication (Steganography)5Introduction 9Some Applications X-ray imaging and radiology Computer Tomography[545x700 24-bit color JPEG, 69069bytes] Section through Visible HumanMale - head, including cerebellum,cerebral cortex, brainstem, nasal passages(from Head subset)http://www.nlm.nih.gov/research/visible/photos.html)Introduction 10An Ultrasound imageProfile of a fetus at four months. This face is approximately 4cmlong. (http://www.parenthood.com)6Introduction 11Computer Tomography Generating 3-D images from 2-D slices. CAD, CAM applications Industrial inspectionsCT Scanner Picker PQ 6000 Model•GE Medical High Speed Advantagescanner•Picker PQ 6000Introduction 12Image/video Processing Methods Image Enhancement Image Restoration Compression Image reconstruction Morphological image processing Feature extraction and recognitioncomputer vision (next quarter, ECE/CS 181b)7Introduction 13Course OutlineIntroduction 14Image EnhancementEnhancement: Improve the visual quality of the image.Eg. Noise removal using median filtering(from http://www.nist.gov/lispix/imlab/noise/shotfc.html)8Introduction 15Image Restoration same as image enhancement, but you haveadditional information concerning the qualitydegradation. Example: removing motion blurin an image of a fast moving object. Matlab site athttp://www.mathworks.com/products/demos/image/deblur_wiener/deblur.htmlIntroduction 16IP methods (cont.) Reconstruction: reconstruction fromprojections. Used in constructing 3D datafrom 2D projections in computer tomography. Image representation using features– Low level representations using color, texture,shape, motion, etc.– High level features for recognitions; e.g., facialfeatures. Recognition and scene understanding9Introduction 17Image Processing, Pattern Recognition, Graphics,and Computer Vision Image Processing– This is about image to image transformation(image coding, enhancement, restoration, etc.)ECE 178, ECE 278a. Computer Graphics: CS 180/280 Pattern Recognition: ECE 277b Computer Vision: ECE 181b/281bIntroduction 18Course Outline Introduction– Chapters 1-2 2-D Linear Systems– Class notes; Sampling andQuantization– Class notes; Ch 2.4 Image Enhancement– Ch. 3, 4 Image and VideoCoding– JPEG and MPEGstandards– Ch. 7,810Introduction 19Course Project: not this year… Why project?– To learn more about applications of image processing andget hands-on experience.– typically, the material (needed) is NOT covered in class -thus requires independent study (ten weeks is too short tocover all interesting topics!.) Previous year projects– Streaming video– Data hiding– JPEG 2000Introduction 20Data HidingDroeshout engraving of William Shakespeare (192x240)A text message (1535 bytes) Steganography is the art and science ofcommunicating in a way which hides the existenceof the communication. In contrast to cryptography,where the "enemy" is allowed to detect, interceptand modify messages without being able to violatecertain security premises guaranteed by acryptosystem, the goal of steganography is to hidemessages inside other "harmless" messages in away that does not allow any "enemy" to evendetect that there is a second secret message present[Markus Kuhn 1995-07-03].11Introduction 21Results of Embedding TextEmbedded imageRecovered message (loss-less) Steganography is the art and science ofcommunicating in a way which hides the existenceof the communication. In contrast to cryptography,where the "enemy" is allowed to detect, interceptand modify messages without being able to violatecertain security premises guaranteed by acryptosystem, the goal of steganography is to hidemessages inside other "harmless" messages in away that does not allow any "enemy" to evendetect that there is a second secret message present[Markus Kuhn 1995-07-03].Compressed image (lossy JPEG 85%)Introduction 22Example: Image in ImageRenoir’s Le Moulin de la Galette (432x320)Airphoto image (216x160)Embedded Embedded and JPEG compression (85%)Recovered signature image12Introduction 23Example: Video in Video(c) Recovered frame (PSNR 35.7dB)(e) Recovered signature frame (PSNR 45.0dB) (d) Signature frame(cm1002.11700, 352x240) (a) Host frame(cm1002.02500, 352x240) (b) Embedded frame(2M bps, 30 f/s PSNR