An Algorithm for Segmenting Elements in an Image

-Write-up defending a hypothetical algorithm

Segmenting patterns and objects in an image is a challenging problem. Several segmentation algorithms are in use, offering distinct advantages of their own. We propose an algorithm that segments elements in an image taking a totally different approach.

The algorithm that we have developed needs more computation and memory compared to the existing algorithms. Segmentation efficiency however is no better. Yet the algorithm has its place in the research as it opens a new paradigm in approaching the segmentation problem. Instead of seeing segmentation from image processing stand point, we choose to look at it in human interaction stand point. The result is that the segmentation reduces to, focused element identification problem. Humans focus on certain portion of the image where the element is concentrated and try to recognise it. Segmented element then need not precisely be an object corresponding to real word but can be the portion of the image that contains it.

The algorithm that implements this change in approach opens a new avenue for research. We hope that, the shift in approach that the algorithm introduces will result in more practical and efficient solutions to the problem.

~rAGU