ImageProcessing2-Fundamental-bwz

1、Chapter 2Digital Image Fundamental2.1 Elements of Visual Perception2.2 Light and the Electromagnetic Spectrum2.3 Image Sensing and Acquisition 2.4 image sampling and quantization2.5 Basic Relationships Between PixelsDigital Image Fundamentals中英名詞對照:Cornea: 角膜Sclera: 鞏膜Choroid: 脈絡膜Retina:視網(wǎng)膜Fovea: 中央

2、凹Blind spot:盲點Rod: 桿狀細胞Cone: 錐狀細胞Iris: 虹膜Lens: 晶狀體Pupil:瞳孔Diagram from /Structure of Human EyesDiagram from /Structure of Human EyesThe RetinaStructure of Human EyesDiagram from /LightThe retinaPhotoreceptor Densitie

3、sStructure of Human EyesDiagrams from /Structure of Human EyesPhotopic (bright-light) vision : vision with cones6million7millionconcentrated in the central portion of the retina (fovea).color receptorshigh resolution in the fovealess sensitive to lightScotopic (dim-light)

4、 vision : vision with rods75 to 150 millionsdistributed over the retinal surfacecolor blindmuch more sensitive to light (night vision)lower resolutionBlind-Spot ExperimentDraw an image similar to that below on a piece of paper (the dot and cross are about 6 inches apart)Close your right eye and focu

5、s on the cross with your left eyeHold the image about 20 inches away from your face and move it slowly towards youThe dot should disappear!Brightness Adaptation & Discriminationn Because digital images are displayed as a discrete set of intensities, the eyes ability to discriminate between diffe

6、rent intensity levels is an important consideration in DIP results.n The human visual system can perceive approximately 1010 different light intensity levelsn Subjective brightness is a logarithm function of the light intensity incident on the eye.n However, at any one time we can only discriminate

7、between a much smaller number brightness adaptation Brightness Adaptation & DiscriminationnFor a given set of conditions, the current sensitivity level of the visual system is called the brightness adaptation level.Brightness Adaptation & Discrimination Seeing is believing? Perceived brightn

8、ess is not a simple function of intensity：Mach Bands(1865)nSimilarly, the perceived intensity of a region is related to the light intensities of the regions surrounding itBrightness Adaptation & Discrimination Brightness Adaptation & Discrimination Another experiment: a piece of paper on a d

9、esk is always white, but can appear totally black when used to shield the eyes while looking directly at a bright skyOur visual systems play lots of interesting tricks on usOptical IllusionsOptical IllusionsOptical IllusionsNo, theyre both the same sizeIs the left center circle bigger?Optical Illusi

10、onsOptical IllusionsOptical IllusionsLight and the Electromagnetic SpectrumLight is just a particular part of the electromagnetic spectrum that can be sensed by the human eyeNewton, 1666Violet, Blue, Green, Yellow, Orange, and RedBlends smoothly into the next.Light and the Electromagnetic SpectrumLi

11、ght and the Electromagnetic SpectrumnVisible light: 0.430.79um (Wavelength )nThe electromagnetic spectrum is split up according to the wavelengths of different forms of energyWhere c is the speed of the light, v is the frequency, and h is the Plancks constanthvEvc,Light and the Electromagnetic Spect

12、rumnThe colours that we perceive are determined by the nature of the light reflected from an objectnFor example, if white light is shone onto a green object most wavelengths are absorbed, while green light is reflected from the objectWhite LightColours AbsorbedGreen LightLight and the Electromagneti

13、c SpectrumThree basic quantities to describe the quality of a chromatic light source:Radiance (radiant flux or radiant power)(Watts: W)（輻射功率）The total amount of energy that flows from the light source.Luminance (luminous flux)(lumen: lm)(光通量）A measure of the amount of energy an observer perceives fr

14、om a light source.Example: Far Infrared RegionBrightnessSubjective descriptor of light perception that is practically impossible to measure.Light and the Electromagnetic SpectrumnIn principle, if a sensor can be developed that is capable of detecting energy radiated by a band of the electromagnetic

15、spectrum, we can image events of interest in that band.nHowever, the wavelength of an electromagnetic wave require to “see” an object must be of the same size as or smaller than the object.nElectromagnetic waves is not the only method for image generation. Such as sound reflection-ultrasonic imagesL

16、ight and the Electromagnetic SpectrumGamma raysMedical ImagingAstronomical ImagingRadiation in nuclear environmentsHard X RaysIndustrial ApplicationsSoft X RaysChest X-Ray (shorter wavelength end)Dental X-Ray (lower energy end)UltravioletMicroscopy ImagingInfrared region:Near-infraredFar-InfraredMic

17、rowaveMicrowave Ovens, Communication, RadarRadio waveAM, FM, TV, and Medical imagingStellar bodies observationtumourinfraredvisibleFamous scientistsImage AcquisitionImages are typically generated by illuminating a scene and absorbing the energy reflected by the objects in that scene.Image Acquisitio

18、nImaging SensorsImage acquisition sensorsuSingle sensoruStrip sensoruSensor arrayImage Sampling and QuantizationnIncoming energy lands on a sensor material responsive to that type of energy and this generates a continuous voltagenTo create a digital image, we need to convert the continuous sensed da

19、ta into digital form.nThis involves two steps: sampling and quantization.Image Sampling and QuantizationImage Sampling and QuantizationLet Z be the set of real integers R the set of real numbersSampling:Partition the xy plane into a grid.The coordinates of the center of each grid being a pair of ele

20、ments from the Cartesian product Z2.The set of all ordered pairs of elements (zi, zj). With zi and zj being integers from Z.Quantization:f is a function that assigns a gray-level value (a real number in R) to each distinct pair of coordinate (x, y).Image Sampling and Quantizationsampledreal imagequa

21、ntizedsampled & quantized連續(xù)圖像 空間離散 幅度量化 數(shù)字圖像Sampling),(crIS,CIcontinuous imagesampled imageTake the average within each square.),(crIS,CIcontinuous imagesampled imageTake the average within each square.Sampling),(crIS,CIcontinuous imagesampled imageTake the average within each square.Sampling16

22、million colors16 colorsQuantizationRepresentation) 1, 1() 1 , 1()0 , 1() 1, 1 () 1 , 1 ()0 , 1 () 1, 0() 1 , 0()0 , 0(),(NMfMfMfNfffNfffjif1, 11 , 10, 11, 11 , 10, 11, 01 , 00, 0NMMMNNaaaaaaaaaARepresentationRepresentation)(BitkNMbk2LBoth spatial and gray level resolutions determine the storage size

23、 of an image (bytes)e.g. spatial resolution: 40 x 40gray level resolution: 64 (log264 = 6 bits/pixel) The number of pixels: 40 x 40 = 1600 pixels The storage size (no compression, no overhead): 1600 x 6 = 9600 bits = 1200 bytes 1.17 KBUsually, the M and N are positive integers, and the number of gra

24、y levels is an integer power of 2:RepresentationSpatial & Intensity Level ResolutionuThe spatial resolution of an image is determined by how sampling was carried out.uSpatial resolution simply refers to the smallest discernable detail in an imageVision specialists will often talk about pixel siz

25、e or pixel per inch(PPI)Graphic designers will talk about dots per inch (DPI)Spatial & Intensity Level ResolutionIntensity level resolution refers to the number of intensity levels used to represent the imageThe more intensity levels used, the finer the level of detail discernable in an imageInt

26、ensity level resolution is usually given in terms of the number of bits used to store each intensity levelNumber of BitsNumber of Intensity LevelsExamples120, 12400, 01, 10, 114160000, 0101, 1111825600110011, 010101011665,5361010101010101010Spatial & Intensity Level ResolutionSpatial & Inten

27、sity Level Resolution1024 * 1024512 * 512256 * 256128 * 12864 * 6432 * 32Spatial & Intensity Level Resolution128 grey levels (7 bpp)64 grey levels (6 bpp)32 grey levels (5 bpp)16 grey levels (4 bpp)8 grey levels (3 bpp)4 grey levels (2 bpp)2 grey levels (1 bpp)256 grey levels (8 bits per pixel)S

28、patial & Intensity Level ResolutionSpatial resolution: Spatial resolution: M*NGray level resolutionGray level resolution：LHow many samples and gray levels are required for a good approximation?Resolution (the degree of discernible detail) of an image depends on sample number and gray level numbe

29、r.i.e. the more these parameters are increased, the closer the digitized array approximates the original image.But: storage & processing requirements increase rapidly as a function of N, M, and kSpatial & Intensity Level ResolutionThe big question with resolution is always: “how much is enou

30、gh?”This all depends on what is in the image and what you would like to do with itKey questions includeDoes the image look aesthetically pleasing?Can you see what you need to see within the image?Spatial & Intensity Level ResolutionThe picture on the right is fine for counting the number of cars

31、, but not for reading the number plateMatlab experimentI = imread(pout.tif);imshow(I); whosName SizeBytes ClassI291x240 69840 uint8 arrayGrand total is 69840 elements using 69840 bytesMatlab experimentclcclear allclose allX= zeros(256,256);for i = 1:256 X(:,i) = i-1;endI3 = mat2gray(X, 0 255); imsho

32、w(I3); imwrite(I3,c:test.tif); colorbarMatlab experimentHigher gray levels, looks better256 Level 64 Level 16 LevelZooming and Shrinking Digital ImagesZooming OversamplingShrinking SubsamplingZooming and Shrinking Digital ImagesZooming: The creation of new pixel locationsThe assignment of gray level

33、s to those new locationsNearest neighbor interpolation (NN)NN produces checkerboard effectNNGray levelZooming and Shrinking Digital ImagesZooming: Bilinear interpolation已知Q12，Q22，Q11，Q21，但是要插值的點為P點，這就要用雙線性插值了，首先在x軸方向上，對R1和R2兩個點進行插值，這個很簡單，然后根據(jù)R1和R2對P點進行插值，這就是所謂的雙線性插值。Zooming and Shrinking Digital Ima

34、gesZooming: Bilinear interpolation首先在x方向進行線性插值，得到然后在y方向進行線性插值，得到Zooming and Shrinking Digital ImagesNumber of pixels not equal to resolutionNN Interpolation(Pixel replication)Bilinear interpolationZooming and Shrinking Digital ImagesShrinking: Similar manner as just described for zooming.Delete the

35、pixelsK=1/3Basic Relationships Between PixelsA pixel p at (x,y) has 2 horizontal and 2 vertical neighbors:(x+1,y), (x-1,y), (x,y+1), (x,y-1)This set of pixels is called the 4-neighbors of p: N4(p)The 4 diagonal neighbors of p are: (ND(p) - (x+1,y+1), (x+1,y-1), (x-1,y+1), (x-1,y-1) - N4(p) + ND(p) N

36、8(p): the 8-neighbors of pDefinitions: f(x,y): digital image Pixels: p,qN4N4PN4N4NDNDPNDNDN8N8N8N8PN8N8N8N8Basic Relationships Between Pixels000000000000000110000000001111100000011111110000011111110000001111110000001111111000001111111110000001111110000000000000SN4(p)QN8(p)PND(p)AdjacencyWe consider

37、three types of adjacency:4-adjacency: two pixels p and q with values from V are 4-adjacent if q is in the set N4(p)8-adjacency : p & q are 8- adjacent if q is in the set N8(p)m-adjacency: p & q with values from V are m-adjacent ifq is in N4(p) orq is in ND(p) and the set N4(p)N4(q) has no pi

38、xels with values from VAdjacencyMixed adjacency is a modification of 8-adjacency and is used to eliminate the multiple path connections that often arise when 8-adjacency is used.100010110100010110100010110V=1 8-adjacencym-adjacencyConnectivityConnectivity between pixels is important:Because it is us

39、ed in establishing boundaries of objects and components of regions in an imageTwo pixels are connected if:uThey are neighbors (i.e. adjacent in some sense - e.g. N4(p), N8(p), )uTheir gray levels satisfy a specified criterion of similarity (e.g. equality, )V is the set of gray-level values used to d

40、efine adjacency (e.g. V=1 for adjacency of pixels of value 1)PathPathPath(通路)u A sequence of adjacent pixels.u For example: a path from pixel p with coordinate (x, y) to pixel q with coordinate (s, t) is defined (x0, y0)，(x1, y1)，(xn, yn)Where (x0, y0) = (x, y)，(xn, yn) = (s, t), (xi, yi) and (xi-1,

41、 yi-1) are adjacent, 1 i n, n is called the length of the path. uIf (x0, y0) = (xn, yn), the path is a closed path(閉合通路閉合通路). uWe can define 4-, 8-, and m-path depending on the type of adjacency. PathConnected Set and RegionDefinitionsDefinitions:uLet S represent a subset of pixels in an image. Two pixel p and q are said to be connected (connected (連通連通) ) in S if there exists a path between them.uFor any pixel p in S, the set of pixels that are connected to it in S