Adaptive multi-scale AMSS operator for quality detection of silks

Abstract

In order to de-noise the silk images corrupted by various noises, a new improved filter based on affine morphological scale space (AMSS) equation was proposed. In this new operator both the adaptive idea and scale space were fused into partial differential equations. Scale parameters were changed adaptively according to the presence or absence of noise in image which was detected by directional gradient operators with four templates. A new evaluating criterion was constructed to test the performance of the improved AMSS operators. Two parameters based on mean square error and minimum absolute error could be acquired to evaluate the effect of keeping edge information and the ability of smoothing noise. Experimental results illustrate that this improved operator can efficiently keep edge information, and the de-noising performance is more effective than other methods for the image with multi-noise.

Keywords:

• image processing
• AMSS
• PDEs
• noise reduction
• quality detection

AMS Subject Classifications :

• 65K05
• 90C30

Acknowledgements

The work is supported by the Zhejiang Provincial Key Fundamental Research and Development Program, China (No. 2006C21037).