New approaches of multifractal image analysis

Some new approaches of multifractal image analysis methods are introduced in this paper. The greatest advantage of multifractal analysis methods is the measure-dependent characteristic. After some existing measures have been addressed, we propose two new measures derived from thresholding using Renyi's entropy for further multifractal analysis. Then we propose two novel image representations based on the multifractal image analysis assumption: the Holder exponent representation and multifractal dimension distribution representation. Experimental results show that even for general images, such representations could be utilized for further image analysis, such as feature extraction and edge detection.