Implementation of segmentation scheme based on wavelet transform in multi-spectral fluctuation patterns

Segmentation is one of signal processing methods that is fruitful to recognize some details of parts contained in a detail pattern before will be further processed. this study proposed a segmentation scheme,which is the input is based on the output of approximation of level two from wavelet transformation of 2d-dwt (two dimension discrete wavelet transform). the segmentation scheme and algorithm process are implemented on the fluctuations patterns of hhf (high high-fluctuation) in multi spectral that previously are extracted. the fluctuation patterns of hhf are 2d (dimensional) graphic that consist of matrix of the average value and standard deviation inside. the method that will be used is to apply the approach of segmentation scheme,which is suitable for treating hhf fluctuation pattern. moreover,it also employs the approach to acquire the highest amplitude value and also to present the signals in the segments that have the top peak these fluctuations. in addition,there are some parameters that are very prominent in segmentation results to be analyzed,such as: the number of segments,the input frequency,spectral noise,the peak of amplitude value and vmr (variance to mean ratio). furthermore,the analysis results of these parameters will be compared between the highest top three segments in the pattern fluctuation of hhf. based on the results,it is noticeable to say that the segmentation scheme can be implemented for hhf pattern. moreover,there are some segments that have the highest peak of amplitude values and also some parameters show quite striking phenomenon compare to other segments with the same parameters. in addition,the result of vmr is suitable with the statistical approach at vmr < 1. it can be said that the proposed segmentation approach can provide the illustration of the segment each of pattern fluctuations hhf clearly.