Histogram Analysis Using a Scale-Space Approach
IEEE Transactions on Pattern Analysis and Machine Intelligence
Region Competition: Unifying Snakes, Region Growing, and Bayes/MDL for Multiband Image Segmentation
IEEE Transactions on Pattern Analysis and Machine Intelligence
Clustering by Scale-Space Filtering
IEEE Transactions on Pattern Analysis and Machine Intelligence
Mean Shift: A Robust Approach Toward Feature Space Analysis
IEEE Transactions on Pattern Analysis and Machine Intelligence
IEEE Transactions on Information Technology in Biomedicine
| Hi-index | 0.00 |
This paper presents a two-stage machine learning method by simulating visual system for segmentation of marrow cell image. Firstly, the scale space clustering is employed to simulate primary visual system to separate image into series regions with similar colours. Different from traditional methods, we focus on a few significant clusters rather than all of them. Priori knowledge is considered to group useful samples for machine learning to simulate visual attention. Secondly, SVM classifier is used to discriminate the pixels of object from background. We could control the performance of classifier by constructing the training set of SVM according to priori knowledge and the characteristics of cell structure. So visual attention could be realized in some degree in our method. Experimental results demonstrate the new method is more accurate and robust than standard SSF (Scale space filter) and mean-shift based algorithm without attention.