Two-Dimensional Stockwell Transform Based Image Fusion for Combining Multifocal Images

In real time image acquisition, especially in the areas of remote sensing and telemedicine, capturing the entire scenario/specimen/scene is not possible even with high resolution sensors. Hence repeated image analysis is done on a large set of images. It increases the computational complexity of the analysis system especially when a large set of images is to be analyzed. Hence it is necessary to identify a technique that reduces the computational time. One such technique is image fusion which allows a set of images to be combined. Image fusion is defined as the process of combining two images which not only retains complexity info but also strengths redundant information. Conventionally wavelet transforms are used or decomposing the images into low and high frequency components. Later choose maximum and choose average are used for combining the coefficients. Though wavelet transfer provides the spectral information, it could not provide details about the phase or phase difference. Also in case of bursts in images, wavelet transform could not provide the appropriate information. Hence it is necessary to identify a suitable transform for obtaining the burst details. Stockwell transform provides frequency, time and phase information. In this work, a set of multi focal images are acquired and Stockwell transform is applied on this image. Later appropriate fusion rules are used to combine this features and inverse Stockwell is used on the fused image to obtain the original image. Performance of the proposed technique is measured in terms of Root Mean Square Error and standard deviation. Performance of the proposed technique is compared with that of wavelets based image fusion and is found that Stockwell transform provides better results (both quantitatively and qualitatively) than wavelet transform.