Infrared and visible image fusion brings complementary information from different images into a single image, improving its information processing capabilities. There is always a quest for maximum information extraction and subsequent transfer to fused images. In this paper, a scale-aware infrared and visible image fusion algorithm based on sparse representation (SR) and weighted least square (WLS) optimization is proposed. Firstly, a guided filter is employed to maximize information extraction from visible images. The source images are decomposed using a rolling guidance filter (RGF) based on the LoG filter and joint bilateral filter. RGF has both scale-aware and edge-preservation properties; it represents the image information at specific levels. The low-frequency coefficients are fused using sparse representation based on a Modified Prewitt-based clustered dictionary and high-frequency coefficients are fused using the max-absolute rule and WLS optimization. This optimization prevents the transfer of distortion and redundant information from the infrared image to the fused image. The subjective and objective performance evaluations show that the proposed algorithm could outperform other state-of-the-art techniques.