Color image fusion using color space transform and support value transform

Abstract

This paper considers image under least squares support vector machine (LS-SVM) framework and introduces support value transform (SVT) into color image fusion and shows how it can be applied to image enhancement: extension of depth of focus. Advantages of color space transform and support value transform are combined for identifying salient features in the source images at multiple scales and integrating those features into a single composite image result. Firstly, the original images are transformed in YUV space and are carried on the support value decomposition to each color component of all images; Secondly, we apply different fusion rules to the corresponding high and low frequency component from the support value decomposition, thereby the support values are constructed; Then, the last fusion support values are generated based on consistence measurement, in which luminance Y component is used as scale standard; Lastly, we carry on support value inverse transform and YUV inverse transform and the fusion image is obtained. The support value analysis is developed by using a series of multi-scale support value filters, which are obtained by filling zeros in the basic support value filter deduced from the mapped LS-SVM to match the resolution of desired level. This proposed fusion framework is then used to combine a set of color source images, taken from a sensor with varying focus settings, into a single fused image result that has improved depth of field over any of the other frames in the input sequence. The results show the performance of the depth-of-field extension on imagery and demonstrate that the proposed approach is effective.