Welcome to ImSegm’s documentation!

Indices and tables

General superpixel image segmentation: (un)supervised, center detection, region growing

This package is aiming at (un/semi)supervised segmentation on superpixels withcomputing some basic colour and texture features. This general segmentationcan be followed by an object centre detection and proximate ellipse fittingto expected object boundaries. Last included method is region growingwith learned shape prior also running on superpixel grid. The package contains several low-level Cython implementation to speed up somefeature extraction methods. Overall the project/repository contains example codes with visualisationin ipython notebooks and experiments required for replicating all published results.

Superpixel segmentation with GraphCut regularisation

Image segmentation is widely used as an initial phase of many image processingtasks in computer vision and image analysis. Many recent segmentation methodsuse superpixels because they reduce the size of the segmentation problemby order of magnitude. Also, features on superpixels are much more robustthan features on pixels only. We use spatial regularisation on superpixelsto make segmented regions more compact. The segmentation pipeline comprises(i) computation of superpixels; (ii) extraction of descriptors such as colourand texture; (iii) soft classification, using a standard classifier for supervisedlearning, or the Gaussian Mixture Model for unsupervised learning; (iv) finalsegmentation using Graph Cut. We use this segmentation pipeline on real-worldapplications in medical imaging. We also show that unsupervised segmentationis sufficient for some situations, and provides similar results to those obtainedusing trained segmentation.

Object centre detection and Ellipse approximation

An image processing pipeline to detect and localize Drosophila egg chambers thatconsists of the following steps: (i) superpixel-based image segmentationinto relevant tissue classes (see above); (ii) detection of egg center candidatesusing label histograms and ray features; (iii) clustering of center candidates and;(iv) area-based maximum likelihood ellipse model fitting.

Superpixel Region Growing with Shape prior

Region growing is a classical image segmentation method based on hierarchicalregion aggregation using local similarity rules. Our proposed approach differsfrom standard region growing in three essential aspects. First, it workson the level of superpixels instead of pixels, which leads to a substantial speedup.Second, our method uses learned statistical shape properties which encourage growingleading to plausible shapes. In particular, we use ray features to describethe object boundary. Third, our method can segment multiple objects and ensurethat the segmentations do not overlap. The problem is represented as energyminimisation and is solved either greedily, or iteratively using GraphCuts.

References

  • Borovec J., Svihlik J., Kybic J., Habart D. (2017). Supervised and unsupervisedsegmentation using superpixels, model estimation, and Graph Cut.SPIE Journal of Electronic Imaging 26(6), 061610. DOI: 10.1117/1.JEI.26.6.061610.

  • Borovec J., Kybic J., Nava R. (2017) Detection and Localization of DrosophilaEgg Chambers in Microscopy Images. In: Wang Q., Shi Y., Suk HI., Suzuki K. (eds)Machine Learning in Medical Imaging. MLMI 2017. LNCS, vol 10541. Springer, Cham.DOI: 10.1007/978-3-319-67389-9_3.

  • Borovec J., Kybic J., Sugimoto, A. (2017). Region growing using superpixelswith learned shape prior. SPIE Journal of Electronic Imaging 26(6), 061611.DOI: 10.1117/1.JEI.26.6.061611.