Image Analysis

Image Analysis: Convolution Masks, ROI (Region Of intrests), arithmetic and logical operations, spatial filtering linear and nonlinear.

Binary image analysis: thresholding, connectivity, labeling algorithm, object properties (area, centroid, eulertal etc.).

Edge Detection: Gradient Operator (sobel, robert, Prewitt), compass masks, LoG (Laplace of Gaussian), Hough transformation.

Segmentation: Splitting and merging (Split and merge), Watershed transform.

Morphological filtering: Structure element, erode, dilate, opening, closing, thinning, etc.

Fourier Transform: FFT interpretation of images.

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).

Object recognition: object description (descriptor), form-based (Fourier descriptor, chain code), region-based (moments, area, perimeter, etc.)

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.

Geometric transformations: transformation matrices. Transformation types (affine, projective), tie points.

Color images: color models (RGB, HSI, HSV, Lab), color scheme, pseudo colors, segmentation based on colors.

Lighting: Lighting function: (contrast, shadow, structure, reflections, edges).

Light Type: (LED, IR, polarization). Light illumination: (Front light, side lights, back light)

Restoration and reconstruction: Degradation: Motion, lens error, CCD array. Inversfilter. Wiener Filter.

Camera calibration: Camera model.

Stereo vision: Stereo matching (SAD,SSD,NCC), disparity map.

Læringsutbytte - Kunnskap:

The candidate should have knowledge about

techniques for image processing

methods for segmentation of images

characterization of objects

methods of classification

   

Læringsutbytte - Ferdigheter:

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 tools

   

Læringsutbytte - Kompetanse:

The candidate should have expertise in

methods of image analysis, object recognition and classification

the various elements in a Machine Vision System

Pedagogiske metoder:

Lecture, exercise and mini project (duration 4 weeks).    

Obligatoriske arbeidskrav:

8 assignments. 80% of computer exercises must be approved. These must be delivered in time andplaced in folders.