BASICS OF DIGITAL IMAGE PROCESSING Erkki Rämö
20. 3. 2018 2 Digital image processing § Editing and interpreting of picture information § Examples: § Improving the visual quality of the image § Removing an error from the image § Automated interpretation of the image
Related disciplines
20. 3. 2018 4 Group discussion 1 • Discuss application areas of digital image processing and analysis.
20. 3. 2018 5 Where is image processing applied? § Biological research – cell studies
20. 3. 2018 Lauri Toivio § Military research – interpretation of reconnaissance photos 6
20. 3. 2018 Lauri Toivio § Document control – scanning, interpretation, archiving 7
20. 3. 2018 Lauri Toivio § Industry automation – machine vision 8
20. 3. 2018 Lauri Toivio § Forensics – Fingerprint analysis 9
20. 3. 2018 Lauri Toivio § Medicine – x-ray image analysis 10
20. 3. 2018 Lauri Toivio § Photography – Digital photography 11
20. 3. 2018 § Publishing 12
20. 3. 2018 Erkki Rämö §Space investigation 13
20. 3. 2018 Erkki Rämö §Remote Sensing 14
20. 3. 2018 Erkki Rämö §Mapping (eg. Google street view) 15
20. 3. 2018 Erkki Rämö §Film industry 16
20. 3. 2018 17 Visual image §Light = electromagnetic radiation §Different wavelengths of light reflect from the object and absorb to the object in different ways, depending on objects surfaces construction and material §Reflecting light is perceived with the eye-brain visual system as an image §Wavelength of visual light is 400 - 700 nm
20. 3. 2018 Perceiving of the visual image What is needed: §Light source § Light bulb radiates light of some color §Target § which reflects part of the light and absorbs the rest §Eye § receives the signal § signal is interpreted by brain 18
Spectrumoflight 10 -6 10 Cosmic Gamma Xrays 103 UV Infrared nm 109 Microwave Radio Visible light 400 nm 700 nm
20. 3. 2018 20 Group discussion 2 • List imaging applications working in different wavelengths. • Can you find imaging using else than electromagnetic radiation
20. 3. 2018 21 Eyesight § Eye, visual nervetrack and brains visual § § centre form the human visual system There’s no visual system better than the eye Some animal eyes are better than human eye
From optical image to a digital image
The construction of the eye Cross-section of the human eye
20. 3. 2018 24 Comparison between an eye and a camera §Similarities: § In the eye image is drawn upside down to the retina § Pupil works like the iris of the camera § Retina, with two types of visual cells, rods (about 120 million) and cones(about 7 million) §Differencies: § Focus by changing the refraction of the lense by means of the radial deformation
20. 3. 2018 25 Visual cells of the eye §Rods § thousands of times more sensitive than cones. § responsible of dark vision §Cones § Responsible of seeing the colors § Three kinds: sensitive for blue-purple, green and redyellow. § Peaks of sensitivity are in the wavelengths of 447 nm, 540 nm ja 577 nm
20. 3. 2018 26 Anatomy of the eye § In the area of accurate sight, in the middle of the yellowspots fovea there are no rods but plenty of cones. § Outside the fovea, accuracy of vision is poor § 5° from the fovea – only half of the accuracy § Only a small area of field of vision is seen accurate § Moving the eyeball we can focus on different details
27 Anatomy of the eye 2 § Sensitivity of visual cells to alteration of lighting is logarithmic § Webers law JND=K*I Where K is constant and JND (Just Noticable Difference) Example: 100 W lighting 10 W power increment. In 1000 W lighting we need 100 W increment for same result § Image: Intensity must be doubled to notice the same visual difference
20. 3. 2018 28 Visual cells react with one another § Mach Band Effect § Eye works like a high pass filter sharpening the details § On the edge of the tone slope, dark color seems lighter and light color seems darker
20. 3. 2018 Influence of the background Simultaneous contrast 29
20. 3. 2018 Influence of the background Simultaneous contrast 30
31 Frequency response § How small details are still visible? § Influences: § Number and positioning of cells, elasticity of the eye, brain response, intensity of light
20. 3. 2018 32 Procedure classes of image processing § Procedures have been developed already in 1960’s, though due to lack of computing power they were hard to implement § Some procedures enhance the quality of the image § Others pick and analyze information from the image
20. 3. 2018 5 Procedure classes 1. Image Enhancement 2. Image Restoration 3. Image Analysis 4. Image Compression 5. Image Synthesis 33
20. 3. 2018 34 Image Enhancement §Most common procedure class § Can be used as independent enhancement method or as pre-operation for other methods, for example reducing the image before analysis
20. 3. 2018 Erkki Rämö 35 Image enhancement 2 §Goal is to enhance the visual quality of the image § contrast and brightness § noise reduction § sharpening
20. 3. 2018 Lauri Toivio Example: adjusting contrast §Photoshop ”autolevels”, which implements the whole tone scale for the image 36
20. 3. 2018 37 Image Restoration §Goal is to restore an image as original or §removal of known photographic error §Corrections: • Removal of geometric distortion • Removal of blur • Noise removal • Motion-blur removal
20. 3. 2018 Example: enhancing sharpness § Photoshop ”Unsharp mask” 38
20. 3. 2018 39 Image Analysis § As result there rarely is an image, but information about what’s in the image § Implemented in various tasks involving artificial vision
20. 3. 2018 Example: Measuring of an object 40
20. 3. 2018 41 Image compression §Goal is to compress image-information so that it would consume less space §Pros § needs less space § faster transfer §Methods: § lossless compression(max 2: 1) § lossy compression(max 100: 1) §Based on redundant information in the image
20. 3. 2018 Lauri Toivio Example: JPEG-compression 183 KB 17 KB 42
20. 3. 2018 Image Synthesis §Image is built out of other images or §Visualization of non-image information §Used when: § taking a picture is not physically possible § fast and/or slow events § modelling an object which does not exist §Examples: § 2 D images of projection images mathematically § visualization of chart information as an image 43
20. 3. 2018 44 Construction of image processing application Application can be divided into unit tasks • Application construction: Applications Fundamental Classes Operations Process
20. 3. 2018 Lauri Toivio 45 Application level § Basic description of application § Example application: § Capture video image of cars licence plate § Process and interpret the signs on the plate § Check register if the vehicle has any offense
20. 3. 2018 46 Image processing part § Process image and identify letters and numbers as an array § In short: Read the signs of the licence plate
20. 3. 2018 47 Process classes § Divide application into unit tasks § Image enhancing: Improve the image quality § Image analysis: Interpret the letters and numbers of the plate ZHO-408
48 Operations § Image enhancement: Improve the image quality § Contrast alteration: steepen the contrast § Edge highlighting: Outlines of signs § Image analysis: Interpretation of the letters and numbers of the plate § Detaching edges: Follow the outlines § Classification of objects: Fit vectors into images in model library
20. 3. 2018 49 Methods §Contrast alteration: steepening of contrast § Contrast stretching as pixel operation §Edge highlighting: Outlines of symbols § Sobels edge highlighting algorithm §Finding edges: Follow the outlines § Edge finding algorithm §Classification of vectors: Fit vectors into images in model library
Скачать презентацию BASICS OF DIGITAL IMAGE PROCESSING Erkki Rämö
9b626cc68c0ed44db392bf63b67473ce.ppt