Multi-Focus Image Fusion- A Technical Review

The imaging equipment usually has difficulty in shooting the target object in which all the objects are effectively in focused. Image fusion plays a vital role in many applications. To overcome it multi-focus image fusion technology has emerged. An image is corrupted by noise blurring or limited focal length or due to different sensors and can have the poor visual quality. Image fusion is used to enhance the quality of a degraded image. It is one of the important task and pre-processing step in digital image processing. Image fusion may be categorized into two broad domains which are Spatial Domain and Transform Domain. There are basically three levels of fusion: 1) Pixel level fusion 2) Feature level fusion and 3) Decision level fusion. The main techniques for pixel level image fusion are average method, principle component analysis, wavelet transform, Brovey transform. For feature level k-means clustering, Region based segmentation and in decision level artificial neural network methods, for gray-scale and RGB images. Earlier proposed method suffers from the noise, artifacts and spectral degradation. The average method leads to the undesirable side effects such as reduced contrast. The weighted wavelet-based method for fusion of PET and CT images has been proposed a pyramid method used for image fusion suffers from blocking artifacts and creates undesired edges. A neural network based fusion requires training sets to get good output.