MULTIFOCUS IMAGE FUSION BASED ON LUCIDITY DECISION PARAMETERS

Abstract

The goal of image fusion is to combine information from multiple images of the same scene. The result of image fusion is a single image which is more suitable for human perception or further image processing tasks. It is obtained by extracting all the useful information from the source images while not introducing artifacts or inconsistencies which can distract human observers or the following processing. Nowadays, image fusion has become an emerging and essential tool and shown its power in many fields like image analysis and computer vision, automatic object detection, robotics, military and law enforcement, satellite imagery, night vision applications, remote sensing and medical diagnosis. For this purpose a novel image fusion technique has been proposed. Firstly original input multi-focus images are partitioned into blocks. Then clarity of these blocks is decided on the basis of three distinctive features i.e. Spatial Frequency, Image Clarity and Block Visibility. Using this decision, the original input multi-focus images are further decomposed into much smaller blocks and again decision is made for blocks which are on the boundary of focused and blurred portions on the basis of the same three distinguishing features. After this practice, all focused and blurred blocks of the original images are clearly identified. Then the smaller blocks which are on the boundary of the clear and blurred parts are fused using conventional wavelet transform and all other blocks away from boundary are taken from original images as intact. Experimental results on standard test images (i.e. Lena, Barbara and Peppers) clearly depict that the proposed approach outshines classical discrete wavelet transform based image fusion techniques and many other. These results provide higher Peak Signal to Noise Ratio (PSNR) and smaller Root Mean Square Error (RMSE) values than some of the previous approaches. MATLAB 7.0 has been used for the implementation of the proposed approach. Experiments have been carried out on a variety of standard grayscale images with different defocus parts.