A novel PCNN with GA based optimized approach for pixel level multimodal image fusion with empirical mode decomposition

Abstract

Multi-modal image fusion is a simple passageway for doctors to perceive the injury to dissect images of distinctive modalities. Distinctive imaging modalities such as Computed Tomography (CT), Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET), Single Photon Emission Tomography (SPECT) etc. provides distinctive information about the human body which is important in diagnosing diseases. However, these modalities are not equipped to provide complete information by observational constraints. Hence the objective of image fusion is to process the content at every pixel position in the input images and sustain the data from that image which represents the genuine scene or upgrades the potency of the fused image for an accurate application. The fused image provides an intuition to data contained within multiple images in a single image which facilitates physicians to diagnose diseases in a more effective manner. Here image fusion has been performed by utilizing five distinct techniques-Discrete Wavelet Transform (DWT), Stationary Wavelet Transform (SWT), Non Subsampled Contourlet Transform (NSCT), Pulse Coupled Neural Network (PCNN) and Pulse Coupled Neural Network (PCNN) with Genetic Algorithm (GA) using Empirical Mode Decomposition (EMD) optimization technique. Quantitative and qualitative analysis illustrates that Pulse Coupled Neural Network (PCNN) with Genetic Algorithm (GA) using Empirical Mode Decomposition (EMD) optimization technique outperforms than other image fusion strategies. newline