Impact of co-efficient selection rules on the performance of DWT based fusion on medical images

CT image is sensitive to bones whereas PET image indicates the brain function with low spatial resolution. This paper focuses on fusion methods for eight sets of PET and CT images based on the Discrete Wavelet Transform (DWT), the most popular tool for image processing. For fusing low frequency coefficients maximum and average rule and for high frequency coefficients contrast, gradient and maximum rules are applied for fusion. On comparing the different fusion results, it could be observed that the best method for low frequency coefficients is average fusion rule and for high frequency coefficients it is gradient fusion rule. From the observation of different fusion rules, entropy and Peak Signal to Noise Ratio values are high whereas Root Mean Square Error and Standard Deviation values get decreased for average-gradient method.