An almost automatic image fusion scheme for balancing clarity and visual effects

Abstract

Multi-focus image fusion aims to combine multiple images with different focuses to form a single sharp image. The basic principle is to first compare the local content information of every block or pixel from distinct input images, and then choose the largest contrast among them because the maximum is generally viewed as the clearest. However, this is usually impossible unless a perfect sharpness measure is adopted. For some cases, especially with smooth and plain areas on images, a higher value of sharpness measurement does not always come from a more focused region and thus a wrong selection of a block or pixel is prone to bring about blocking effects. In addition, how to select the optimal block size suitable for all sorts of images is another relatively important issue, but so far no existing method has provided the skill. On the other hand, modern techniques tend to obtain an effective algorithm by some complicated procedures, thereby increasing computational time and being difficult to implement. In order to effectively overcome potential blocking effects and provide a simple as well as easily extended approach, an almost automatic image fusion scheme for balancing clarity and visual effects, a combination of a sharpness measure and the occurrence rate of the maximum sharpness for each pixel lying in different block sizes, is proposed. Through the proposed scheme, three regions are naturally obtained: the focus region, the transition region and the blur region. Experimental results confirm that the proposed scheme does efficiently achieve much satisfactory visual quality.