02 –Digital Image Fundamentals - Jimmy The Lecturer

1/25/201702 – Digital Image Fundamentalsprepared by 4sCVe1

1/25/2017Image Processing ProblemsThere are many image processing applications andproblems, we will consider the following basic classes ofproblems Image Representation & Modeling Image Enhancement Image Restoration Image Analysis Image Reconstruction Image Data CompressionImage Representation & ModelingIn image representation one is concerned withcharacterization of the quantity that each picture element(pixel) represents. represent luminance of objects in a scene (taken by ordinarycamera) represent the absorption characteristics of the body tissue (X‐rayimaging) represent radar cross section of a target (radar imaging) represent the temperature profile of a region (infrared imaging) represent the gravitational field in an area (geophysical imaging)2

1/25/2017Image representation and modelingImage EnhancementThe goal is to accentuate certain image features forsubsequent analysis or for image display. Contrast and Edge enhancementPseudo coloringNoise filteringSharpeningMagnifyingImage enhancement is useful in featureextraction, image analysis, and visualinformation display3

1/25/2017Image RestorationImage restoration refers to removal or minimization ofknown degradations in an image. This includes deblurringof images degraded by the limitations of a sensor or itsenvironment, noise filtering, and correction of geometricdistortion or non‐linearities due to sensors.The image of an object can be expressed asThe imagepoint spread The objectfunctionadditive noisefunctionImage Analysis Image analysis is concerned with making quantitativemeasurements from an image to produce a descriptionof it.Image analysis techniques require extraction of certainfeatures that aid in the identification of the objectSegmentation technique are used to isolate the desiredobject from the scene so that the measurements can bemade on it subsequently.Quantitative measurements of object features allowclassification and description of the image4

1/25/2017Image Reconstruction Image reconstruction from projections is a special classof image restoration problems where a two‐ (or higher)dimensional object is reconstructed from several one‐dimensional projections.Planar projections are thus obtained by viewing theobject from many different angles.Image Reconstruction5

1/25/2017Image Data CompressionThe amount of data associated with visual information is solarge that its storage would require enormous storagecapacity.Image Data CompressionStorage and/or transmission of such data require largecapacity and /or bandwidth, which could be very expensive.Image data compression techniques are concerned withreduction of the number of bits required to store ortransmit images without any appreciable loss ofinformation6

1/25/2017Image Sampling and QuantizationTo create a digital image, we need to convert thecontinuous sensed data into digital form.continuous with respect to thex‐ and y‐coordinates, and also inamplitudeSamplingDigitizing thecoordinate valuesQuantizationDigitizing theamplitude values7

1/25/2017Image Sampling and QuantizationContinuous imageDigital imageMrowsimage f(x, y)NcolumnsThe result of sampling andquantization is a matrix ofreal numbersCoordinate ConventionEach element of this matrixarray is called an imageelement, picture element,pixel, or pel8

1/25/2017Convention in MATLABDigital image represented in MATLAB matrixBlack &WhiteGrayscaleColouredImageThe number of colours in an image depends onthe colour depth (the number of bits per pixel9

1/25/2017Colour DepthThe number of distinct colours that can be represented by apixel depends on the number of bits per pixel (bpp)Gray LevelColours in imageColour Depth (Gray Level) 1 bpp 2 colours (Black and White)2 bpp 4 colours3 bpp 8 colours8 bpp 256 colours16 bpp 65,536 colours (“Highcolor”)24 bpp 16.7 million colours (“Truecolor”)Storage Requirements10

1/25/2017Storage RequirementsLet the resolution of a digital image be 256 x 256. It is an 8bit grayscale image. How many KBs are required to storethis image?Size (Storage Capacity) 256 x 256 x 8 524,288 bits 65,536 bytes 64 KBBlack and White Images Binary ImageIntensity of Black 0Intensity of White 111

1/25/2017Grayscale Image 8 bit grayscale image has256 pixel intensitiesRange : 0 to 255Intensity of Black 0Intensity of Gray 127Intensity of White 255Colour Images Different models used for colour imagesRGB : combination of 3 planes (channels)In MATLAB, 3 different 2‐D arrays, each for Red, Green,and Blue12

1/25/2017Colour ImagesColour Image to Grayscale ImageGrayscale (R G B) / 313

1/25/2017Colour Image to Grayscale ImageWeighted ConversionGrayscale (0.3 * R 0.59 * G 0.11 * B)Different Type of ImagesSame picture in three different modes14

1/25/201715

02 –Digital Image Fundamentals prepared by jimmyhasugian. 1/25/2017 2 There are many image processing applications and . To create a digital image, we need to convert the continuoussensed data into digital form. Image Sampling and Quantization Sampling Quanti