1 Image Processing(IP) 1. Introduction 2. Digital Image Fundamentals 3. Image Enhancement in the spatial Domain 4. Image Enhancement in the Frequency Domain.

11

Image Processing(IP)Image Processing(IP)

1. 1. IntroductionIntroduction2. Digital Image Fundamentals2. Digital Image Fundamentals3. Image Enhancement in the spatial Domain3. Image Enhancement in the spatial Domain4. Image Enhancement in the Frequency Domain4. Image Enhancement in the Frequency Domain5. Image Restoration5. Image Restoration6. Color Image Processing6. Color Image Processing7. Wavelets and Multiresolution Processing7. Wavelets and Multiresolution Processing8. Image Compression8. Image Compression9. Morphological Image Processing 9. Morphological Image Processing 10. Image Segmentation10. Image Segmentation11. Representation & Description11. Representation & Description12. Object Recognition 12. Object Recognition

22

IntroductionIntroduction

1.1 1.1 What is Digital Image ProcessingWhat is Digital Image Processing

1.2 The Origins of Digital Image Processing1.2 The Origins of Digital Image Processing

1.3 Examples of Field that Use Digital Image 1.3 Examples of Field that Use Digital Image ProcessingProcessing

1.4 Fundamental Steps in Digital Image Processing1.4 Fundamental Steps in Digital Image Processing

1.5 Components of an Image Processing System1.5 Components of an Image Processing System

1.6 Importance Academic IP Journals Research1.6 Importance Academic IP Journals Research

1.7 Course Requirements1.7 Course Requirements

33

1.1.1 What is Digital Image 1 What is Digital Image ProcessingProcessing

1. 1. Related Terminologies Related Terminologies – a. image ---- stilla. image ---- still– b. picture --- imageb. picture --- image– c. graph ----- conceptualc. graph ----- conceptual– d. pattern --- conceptuald. pattern --- conceptual– e. graphics -- drawingse. graphics -- drawings– f. animation - dynamic graphicsf. animation - dynamic graphics– g. video ------ dynamic imagesg. video ------ dynamic images

44


1. 1. Image ( monochrome image )Image ( monochrome image )– 2-D light intensity function f(x,y)2-D light intensity function f(x,y)– where (x,y): spatial coordinates;where (x,y): spatial coordinates;– value of f : brightness of gray level at (x,y)value of f : brightness of gray level at (x,y)

2. Digital Image 2. Digital Image – image discretized both in spatial and gray image discretized both in spatial and gray

levelslevels

3. Image Elements3. Image Elements– picture elements (pixels or pels)picture elements (pixels or pels)

55

1.1.1 What is Digital Image Processing1 What is Digital Image Processing

4. Related Fields4. Related Fields– a. computer vision (CV) ----------- 3-D IPa. computer vision (CV) ----------- 3-D IP– b. signal processing (SP) ---------- 1-D IPb. signal processing (SP) ---------- 1-D IP– c. computer graphics (CG) -------- generation of drawingsc. computer graphics (CG) -------- generation of drawings– d. image synthesis (IS) ------------ generation of images (IP d. image synthesis (IS) ------------ generation of images (IP

+ CG )+ CG )– e. pattern recognition (PR) ------- theorye. pattern recognition (PR) ------- theory– f. scientific visualization (SV) --- application of ISf. scientific visualization (SV) --- application of IS– g. multimedia technologies ------- application of a thru fg. multimedia technologies ------- application of a thru f

66

Three types of computerize processesThree types of computerize processes ：：– Low-level processesLow-level processes ：：

Primitive operations such as Primitive operations such as ：： image processing to reduce noise, image processing to reduce noise, contrast enhancement, and image sharpening.contrast enhancement, and image sharpening.Both its inputs and outputs are imagesBoth its inputs and outputs are images

– Mid-level processesMid-level processes ：：Segmentation ( partitioning an image into regions or objects)Segmentation ( partitioning an image into regions or objects)Description of those objects to reduce them to a form suitable for Description of those objects to reduce them to a form suitable for computer processing, computer processing, Classification ( recognition) of individual objects.Classification ( recognition) of individual objects.Its inputs generally are images, but its outputs are attributes Its inputs generally are images, but its outputs are attributes extracted form those imageextracted form those image

– High-level processesHigh-level processes ：：““Making sense” of an ensemble of recognized objectsMaking sense” of an ensemble of recognized objects


77

1.2 1.2 The Origins of Digital Image The Origins of Digital Image ProcessingProcessing

1. 1. Improving digitized Improving digitized newspaper in 1920s to newspaper in 1920s to 1950s1950s

88


2. Improving images from space programs 2. Improving images from space programs from 1964from 1964

99


3. From 1960s till now, the IP field has grown 3. From 1960s till now, the IP field has grown vigorouslyvigorously

4. Computer tomography(CT)4. Computer tomography(CT)– an important achievement of in medicine ( has an important achievement of in medicine ( has

won a Nobel Prize)won a Nobel Prize)

1010

1.3 Examples of Field that Use Digital 1.3 Examples of Field that Use Digital Image ProcessingImage Processing

1.3.1 Gamma –1.3.1 Gamma –Ray ImagingRay Imaging

1111


1.3.2 X-ray 1.3.2 X-ray imagingimaging

1212


1.3.3 1.3.3 Imaging in Imaging in the the Ultraviolet Ultraviolet BandBand

1313


1.3.4 Imaging 1.3.4 Imaging in the Visible in the Visible and Infrared and Infrared BandsBands

1414


1.3.4 Imaging in the Visible and Infrared Bands1.3.4 Imaging in the Visible and Infrared Bands

1515



1616



1717


1.3.4 1.3.4 Imaging in Imaging in the Visible the Visible and Infrared and Infrared BandsBands

1818



1919



2020



2121


1.3.5 Imaging in the Microwave Bands1.3.5 Imaging in the Microwave Bands

2222


1.3.6 Imaging in the Radio Bands1.3.6 Imaging in the Radio Bands

2323


1.3.6 Imaging in the Radio Bands1.3.6 Imaging in the Radio Bands

2424


1.3.7 Examples in which Other Imaging Modalities Are Used1.3.7 Examples in which Other Imaging Modalities Are Used

2525



2626



2727



2828

1.4 Fundamental Steps in Digital 1.4 Fundamental Steps in Digital Image ProcessingImage Processing

2929

1.5 Components of an Image 1.5 Components of an Image Processing SystemProcessing System

3030

1.1.6 Important Academic IP 6 Important Academic IP JournalsJournals

1. 1. IEEE Transactions on Pattern Analysis. & Mach. IEEE Transactions on Pattern Analysis. & Mach. IntelligenceIntelligence

2. IEEE Transactions on Systems, Man, and Cybernetics2. IEEE Transactions on Systems, Man, and Cybernetics

3. IEEE Transaction of Image Processing3. IEEE Transaction of Image Processing

4. Computer Vision, Graphics, and Image Processing4. Computer Vision, Graphics, and Image Processing

5. Pattern Recognition5. Pattern Recognition

6. Image and Vision Computing6. Image and Vision Computing

7. International Journal of Computer Vision7. International Journal of Computer Vision

8. Machine Vision and Applications8. Machine Vision and Applications

9. Pattern Recognition Letters9. Pattern Recognition Letters

3131

1.1.7 Course Requirements7 Course Requirements

1. 1. Textbook:Textbook:– R.C.Gonzalez and R.E. Woods, Digital Image Processing, R.C.Gonzalez and R.E. Woods, Digital Image Processing,

Addison-Wesley Pub. Co., Readings Massachusetts, USA,Addison-Wesley Pub. Co., Readings Massachusetts, USA,

2. Grade Evaluation:2. Grade Evaluation:– b. one or two examsb. one or two exams– a. about 3~4 homeworksa. about 3~4 homeworks

3. Pre-requistes3. Pre-requistes– Ability of programing or Experience of IP Software Ability of programing or Experience of IP Software

( MATLAB).( MATLAB).

3232

2.4 2.4 Some Basic Some Basic Relations Between PixelsRelations Between Pixels

2.4.1 2.4.1 Neighborhood of a PixelNeighborhood of a Pixel– 1. Given a pixel p in the center of 91. Given a pixel p in the center of 9– pixels:pixels:

aa bb cc

dd pp ee

ff gg hh

– thenthen– 4-neighbors of p = b, d, e, g;4-neighbors of p = b, d, e, g;– 8-neighbors of p = a, b, c, d, e, f, g, h;8-neighbors of p = a, b, c, d, e, f, g, h;– diagonal neighbors of p = a, c, f, h;diagonal neighbors of p = a, c, f, h;

3333


2.4.2 2.4.2 ConnectivityConnectivity–1. Terms: 4-connected , 8-connected , 4-adjacent , 1. Terms: 4-connected , 8-connected , 4-adjacent ,

8-adjacent , path connected.8-adjacent , path connected.–2. Connected component (c. c.): 2. Connected component (c. c.):

for any pixel p in a set of pixels S , the set of pixels MS for any pixel p in a set of pixels S , the set of pixels MS that are connected to p is a c.c. of p that are connected to p is a c.c. of p

Sc.c.

3434


2.4.3 2.4.3 Labeling of Connected ComponentsLabeling of Connected Components– 1.Gives a pixel p with r and t as its upper and left-hand 1.Gives a pixel p with r and t as its upper and left-hand

neighbors as follows:neighbors as follows: rr

tt pp

– then the following algorithm labels all c.c. in an binary then the following algorithm labels all c.c. in an binary image ( This algorithm for 4-connected)image ( This algorithm for 4-connected)

a. Scan the image from left to right and from top to bottom;a. Scan the image from left to right and from top to bottom;

b. If p = 0 , continue the scan;b. If p = 0 , continue the scan;

c. If p = 1 , exam r and t;c. If p = 1 , exam r and t;

if r = t = 0 assign a new label to p;if r = t = 0 assign a new label to p;

3535


if r = 1 & t = 0 , assign the label of r to p;if r = 1 & t = 0 , assign the label of r to p;

if r = 0 & t = 1 , assign the label of t to p;if r = 0 & t = 1 , assign the label of t to p;

if r = t = 1 & labels of r & t identical, then assign that label to if r = t = 1 & labels of r & t identical, then assign that label to p;p;

if r = t = 1 & labels of r & t different, then assign one of the if r = t = 1 & labels of r & t different, then assign one of the labels to p and make the two labels “equivalent”;labels to p and make the two labels “equivalent”;

d. Sort all the equivalent label pairs into equivalent classes, and d. Sort all the equivalent label pairs into equivalent classes, and assign a distinct label to each class.assign a distinct label to each class.

– 2.Do a second scan thru the image and replace each label by 2.Do a second scan thru the image and replace each label by the label assigned to its equivalent class.the label assigned to its equivalent class.

– 3. For sorting of equivalent labels ,see Section 2.4.43. For sorting of equivalent labels ,see Section 2.4.4– *P.S. The above is for 4-connectivity, another algorithm in *P.S. The above is for 4-connectivity, another algorithm in

textbook for 8-connectivity; textbook for 8-connectivity;

3636


2.4.4 2.4.4 Relations , Equivalence , and Transitive Relations , Equivalence , and Transitive ClosuresClosures– 1. A property :1. A property :– If R is an equivalent relation on a set A , If R is an equivalent relation on a set A ,

then A can be divided into a group of disjoint then A can be divided into a group of disjoint subsets , called equivalent classes , such that aRb subsets , called equivalent classes , such that aRb iff a and b are in the same subset.iff a and b are in the same subset.

3737


2.4.5 2.4.5 Distance MeasuresDistance Measures– 1. Given three pixels p, q, and z with coordinates (x, y), (s, t), 1. Given three pixels p, q, and z with coordinates (x, y), (s, t),

(u, v), respectively, we have the following three types of (u, v), respectively, we have the following three types of distances:distances:

a. Euclidean distance a. Euclidean distance

DDee(p, q) = [(x-s)(p, q) = [(x-s)22+(y-t)+(y-t)22]]1/21/2

b. City-block distanceb. City-block distance

DD44(p, q) = | x - s | + | y - t |(p, q) = | x - s | + | y - t |

– The pixels with DThe pixels with D44=1 to a pixel p are the 4-neighbors of p=1 to a pixel p are the 4-neighbors of p

c. Chessboard distance c. Chessboard distance

DD88(p, q) = max ( | x - s | , | y - t | )(p, q) = max ( | x - s | , | y - t | )

– The pixels with DThe pixels with D88=1 to a pixel p are the 8-neighbors of p=1 to a pixel p are the 8-neighbors of p

3838


2.4.6 2.4.6 Arithmetic/Logic OperationsArithmetic/Logic Operations– 1. Mask operations1. Mask operations

are arithmetic/logic operation applied to the neighborhood are arithmetic/logic operation applied to the neighborhood ( with g. l. z1, z2, …….., z9) of a pixel with g. l. z5, e.g., ( with g. l. z1, z2, …….., z9) of a pixel with g. l. z5, e.g.,

z5z5 z=(z1+z2+z3…….+z9)/9 z=(z1+z2+z3…….+z9)/9

– 2.Notes:2.Notes:a. g. l. = gray level;a. g. l. = gray level;

b. Masks are also called templates, windows, filters, etc.b. Masks are also called templates, windows, filters, etc.

2.5 Image Geometry ( see the textbook)2.5 Image Geometry ( see the textbook)

2.6 Photographic Films ( see the textbook).2.6 Photographic Films ( see the textbook).

1 Image Processing(IP) 1. Introduction 2. Digital Image Fundamentals 3. Image Enhancement in the spatial Domain 4. Image Enhancement in the Frequency Domain.

Documents

digital image image

image monochrome image

digital image processing

image processing system

color image processing

morphological image

image enhancement

image segmentation