0x010088F6312065A70A44B630A6848498EAD80093CFEAFE161DC94D83C7C5C9A5A748B8

In this part, course evaluation process of the Image Analysis course in Autumn 2020, will be introduced. <h2>Procedure</h2> In order to perform the course evaluation, the students select two students to form the reference group, to represent them in the course evaluation meetings.  The class group is 17 students, but not many students accepted to be in the reference group.  However, the students in the course have their own closed group on Facebook to discuss the course and the class group is small, then it was enough to have a reference group with only two students. Since the class group is small, we have a short discussion in the class weekly. These short meetings provide a good opportunity for students to tell their opinions early in the course, to help the teacher to improve and fix the problems early in the course. We have a final meeting (using ZOOM) at the end of the semester (08.12.2020) to discuss the whole course (course materials, lectures, teaching methods, and the evaluation method).    <h2>Feedback from reference group</h2> Relevance: The students find the course relevant to their field of study and useful overall.  Feedback: There were many meetings where we discussed all the issues (in the class). Then, there were several improvements during the course. The course was well structured and clear what we were going to do week by week. The students were satisfied with class, course materials, and teaching method. Some students wanted shorter time to submit the assignments (to not delay the work), and some students prefer project.based assessment.  Teaching method: There were uploaded notes in BB and reference to the book. The course is running in two days, Mondays and Wednesdays. Digital lecturing on Mondays, while using L165 on Wednesdays). The time is divided between normal lectures and practical assignments (labs). We have also a mini project with 25-40% of the course load. The students worked in groups in the project and some of the groups defined their ideas for the project also.  Assignments:  we have 9 lab assignments and a min project that the students must pass in order to get access to the final exam. The labs were designed to practice the theory by using different tools such as: MATLAB, Python, CVIPtools and DIPimage toolbox.  The student can select tool to solve the labs or to implement the mini project. Many students selected python to implement the project. Some students had difficulties in the assignments, but the teacher went through all the assignments in details (one every week).  Tools: The labs are originally written to be solved in MATLAB, Python, CVIPtools and DIPimage toolbox. Some of the students used Python to implement the project task.   <h2>Attendance</h2> The attendance wasn’t good in Mondays because the late time (14:15 – 18:00) and conflict with other courses, but it was very good in Wednesdays (08:15 – 12:00). The lectures were recorded on Mondays. <h2>Exam results</h2> The final exam was an oral exam. The grades are shown in Table I. All the students must pass at least 80% of the mandatory assignments, in order to get access to the exam. All the students got access to the final exam. Only one student didn’t pass the course. The grades were good distributed as following: <ul><li>A (0)</li><li>B (4)</li><li>C (8)</li><li>D (4)</li><li>E (1)</li><li>F (0)</li></ul><h1> </h1><h2>Conclusion </h2> The course was successful in achieving the learning outcomes. The results are good and the students developed their skills in the field. I have found that challenging the students to work individually on small practical assignments and in groups on real world problem (mini project), is very effective method in achieving the learning outcomes. The students applied their knowledge from other courses in the study program in this course, and they found the useful link between different courses (e.g. Signal processing, Real time programming and intelligent systems).

In this part, course description of the Image Analysis course in autumn 2020, will be introduced. <h2>Course Objectives</h2> This course is an introduction to the fundamental concepts and techniques in image processing and analysis, and their applications to solve real life problems. The topics covered include Digital Image Fundamentals, Image analysis, Image Transforms, Image Enhancement, Restoration and Compression, Morphological Image Processing, Fourier Transform, Feature Extraction, Camera Calibration, Application examples are also included. Upon completion of this course, students will be familiar with basic image analysis techniques for solving real problems. Student will also have sufficient expertise in both the theory of image analysis and its wide range of applications. <h2>Topics</h2> The course topics are: 1.    Image Analysis: Convolution Masks, ROI (Region Of intrests), arithmetic and logical operations, spatial filtering linear and nonlinear.2.    Binary image analysis: thresholding, connectivity, labeling algorithm, object properties (area, centroid, eulertal etc.).3.    Edge Detection: Gradient Operator (sobel, robert, Prewitt), compass masks, LoG (Laplace of Gaussian), Hough transformation.4.    Segmentation: Splitting and merging (Split and merge), Watershed transform.5.    Morphological filtering: Structure element, erode, dilate, opening, closing, thinning, etc.6.    Fourier Transform: FFT interpretation of images.7.    Feature Extraction: RST-invariant (rotation, size, translation), histogram type (mean, standard deviation, skew transparency, energy, entropy), form type (moment-based), spectral -(sector and ring-based power), texture-type (based on co -occurance matrix and the law's textures mesh).8.    Object recognition: object description (descriptor), form-based (Fourier descriptor, chain code), region-based (moments, area, perimeter, etc.)9.    Pattern Recognition: Scatterdiagram, scale, distance measurement, minimum distance classifier, k-nearest neighbor, optimal statistical classification (Bayes) trainingset center, testset, learning curves, error classification confusion matrix.10. Geometric transformations: transformation matrices. Transformation types (affine, projective), tie points.11. Color images: color models (RGB, HSI, HSV, Lab), color scheme, pseudo colors, segmentation based on colors.12. Lighting: Lighting function: (contrast, shadow, structure, reflections, edges).13. Light Type: (LED, IR, polarization). Light illumination: (Front light, side lights, back light)14. Restoration and reconstruction: Degradation: Motion, lens error, CCD array. Inversfilter. Wiener Filter.15. Camera calibration: Camera model.16. Stereo vision: Stereo matching (SAD,SSD,NCC), disparity map. <h2>Learning outcome</h2> Knowledge:     The candidate should have knowledge about:·     Techniques for image processing·     Methods for segmentation of images·     Characterization of objects·     Methods of classificationSkills:    The candidate should:·     Have skills in Blob (Binary Large Object) analysis·     Be able to select appropriate feature for identification of object·     Identify, recognize and classify objects with computer toolsCompetence:     The candidate should have expertise in:·     Methods of image analysis, object recognition and classification·     The various elements in a Machine Vision System <h2>Curriculum</h2> The book listed below will be followed during the course. <ul><li>Digital Image Processing and Analysis, 3. utgåve., CRC Press 2018, Scott E. Umbaugh.</li><li> ISBN= 978-1-4987-6602-9.</li></ul><h2> </h2> <h1> </h1>

·     Lectures (online and physical), discussion at group and class level. ·     Discussion board in Blackboard. ·     Lab exercises and mini project. ·     8 Assignments. 80% of computer exercises must be approved. These must be delivered in time and placed in folders.

Oral examination based on exercises and project.80% of lab exercises and the mini project must be approved.

NTNU:TIA:REST

Department of ICT and Natural Sciences

Institutt for IKT og realfag

Spesifiserte trykte og håndskrevne hjelpemidler tillatt. Ingen kalkulator tillatt.

Engelsk

Norsk

Mer om vurdering

Muntlig eksamen basert på mappe og prosjekt.

Faglig innhold

Bildeanalyse: Foldingsmasker, ROI (Region Of Intrest), aritmetriske og logiske operasjonar, romleg filtrering lineær og ulineær.

Bineær bildeanalyse: Terskling, konnektivitet, labeling algoritme, objekt egenskapar (areal, centroide, eulertal etc.).

Kantdeteksjon: Gradient operatorar (sobel, robert, prewitt), kompassmasker, LoG (Laplace of Gaussian), Hough Transformasjonen.

Segmentering: Oppdeling og sammensmelting (Split and merge), vannbasseng transformasjonen (Watershed transform).

Morfologisk filtrering: Strukturelement, erosjon, groing, åpning, lukking,fortynning,etc.

Fourier Transformasjon: FFT tolking av bilder.

Objekt egenskapar (Feature Extraction): RST- invariant (rotasjon, storleik, translasjon), histogramtype (middelverdi, standardavvik, skjevheit, energi, entropi), formtype (momentbaserte), spektraltype ( sektor og ring basert effekt), texturtype ( basert på co-occurance matrix og law¿s textur masker).

Objekt gjennkjenning: Objekt beskrivelse (descriptor), formbasert (Fourier descriptor, kjedekode), regionbasert (moment, areal, omkrets, etc.)

Mønstergjennkjenning: Spredninsdiagram, skalering, avstandsmål,minimum distanse klassifiserer, k-næraste nabo, optimale statistiske klassifiserer (Bayes),treningsset, testset, læringskurver, feilklassifisering, forvirringsmatrise. 

Geometriske transformasjonar: Transformasjons matriser. Transformasjonstypar (affine, projektive), knytningspunkt.

Fargebilder: Fargemodellar (RGB,HSI,HSV,Lab), fargeplan, pseudofargar, segmentering basert på fargar.

Belysning: Lyssettingens funksjon:( kontrast,skygge, struktur, gjennskinn, kanter).

Lystyper: (LED, IR, polarisasjon). Lysoppsetting: (Frontlys, underlys, sidelys)

Restaurering og rekonstruksjon: Degradering: Bevegelse, linsefeil, CCD matrise. Inversfilter. Wienerfilter.

Kamera kalibrering: Kamera modell.

Stereosyn: Stereomatching (SAD,SSD,NCC), disparity map.

Kursmateriell

Obligatorisk:

Scott E Umbaugh: Digital Image Processing and Analysis., CRC Press (2018), ISBN: 978- 1- 4987- 6602- 9

Læringsformer og aktiviteter

Pedagogiske metoder:

Forelesning, dataøvinger og miniprosjekt (varighet 4 veker)Dataøvinger basert på Matlab.    

Obligatoriske arbeidskrav:

8 dataøvinger. 80% av dataøvingene må vere godkjent. Disse skal leveres i rett tid og plasseres i mapper.

Læringsutbytte

Læringsutbytte - Kunnskap:

teknikker for prosessering av bilder

metoder for segmentering av bilder

karakterisering av objekter

metoder for klassifisering

   

Læringsutbytte - Ferdigheter:

ha ferdigheter i Blob (Binary Large OBject) analyse

kunne velge relevante størrelser for identifisering av objekt

kunne identifisere, gjenkjenne og klassifisere objekt med dataverktøy

   

Læringsutbytte - Kompetanse:

metoder for bildeanalyse, objekt- gjenkjenning og klassifisering

de ulike elementer som inngår i et Machine Vision system

Obligatorisk arbeidskrav

Bildeanalyse

Ingeniør

Bachelor i ingeniørfag - Automatiseringsteknikk

Bachelor i ingeniørfag - Data

Web/Lists(guid'8b698a2d-6ad0-4f7d-b3be-219cc1979f02')/Items(7952)/FieldValuesAsText

Bildeanalyse - IE303512