Objectives
- Read in a set of images / photos.
- Interpret training data provided with each image.
- Produce images with segmented labels.
Approach Taken
- When implementing this segmentation system the method chosen to best extract meaning from a subsection, was to calculate a single Gaussian function that best represented the luminance levels.
- The classification of segments over the images was based both on Gaussian properties of each segment and the pixel location within the photo.
- The two classification techniques, previously mentioned, were be combined using Bayesian inference, so the likelihood of a segment label can be determined.
Learning Outcomes
From this project, I had got a general overview of how image processing can be carried out.