MSER

The MSER (Maximally Stable Extremal Regions) module is used to segment an image into those parts that are maximally stable in the specified channel. I.e. the module finds those blobs that don't change much in size while growing the blob to include its surrounding neighbors. The technique starts with each pixel and slowly grows blobs by increasing (MSER Plus) or decreasing (MSER Minus) the intensity range that makes up a particular blob. While this growing of blobs happens the module monitors the growth of each blob and isolates those that don't grow much over a number of intensity increases. This attribute will tend to isolate blobs that are well bounded from their surroundings.

This form of segmentation results in similar results to the Adaptive Threshold module but can perform better on text, labels, etc. as a precursor to OCR (Optical Character Recognition).

Interface

Instructions

1. Min Blob Size - The Minimum size a blob needs to be before being considered an MSER.

2. Max Blob Size - The Maximum size a blob can be to be considered an MSER.

3. Stability - Defines how stable in intensity (i.e. how well defined) a blob needs to be in order to be identified.

4. MSER Plus - Specifies increasing intensity levels are using during blob growth. Isolates darker blobs.

5. MSER Minus - Specifies decreasing intensity levels are using during blob growth. Isolates lighter blobs.

6. Channel - Sometimes isolating this blob growth to a specific color channel can help segmentation. By clicking on the selected channel only that channel will be used during blob growth.

Example

Source	MSER

See Also

Adaptive Threshold

For more information

Wikipedia MSER

New Post

MSER Related Forum Posts	Last post	Posts	Views
None