Remote sensing image fusion is a crucial process that significantly enhances the quality of remote sensing data by integrating multispectral and panchromatic images. However, this integration poses challenges in both spectral and spatial scales. Traditional fusion methods, such as Intensity-Hue-Saturation (IHS) transformation for Component Substitution (CS) and wavelet transform fusion for Multi-Resolution Analysis (MRA), each have their own advantages and disadvantages when striving to generate high-resolution multispectral images. The former enhances spatial resolution significantly but leads to noticeable spectral distortion, while the latter maintains spectral information well but has limited spatial resolution enhancement. To fully leverage the strengths of both types of methods, we introduce a novel fusion method that combines spectral scale detail and spatial scale detail. The proposed method consists of three stages. The first stage is the multispectral image enhancement preprocessing, which takes into consideration the different demands for spectral and spatial information of different features in remote sensing images. The panchromatic image is used as a guiding image, and guided filtering is applied to enhance the multispectral image. After enhancement, the texture-rich regions in the multispectral image are sharpened, the gradient's information of which are increased. Meanwhile the spectral-rich regions are smoothed with average filtering to preserve spectral information. The enhanced multispectral image is the foundation for controlling the amount of detail injection in the subsequent stage. The second stage is the detail injection stage, which is a key component of the proposed method and includes detail extraction, detail injection coefficient calculation, and spectral preservation coefficient calculation. Initially, spectral scale difference details and spatial scale difference details are extracted separately from the multispectral and panchromatic images using classical component substitution methods and multi-scale analysis methods. Since there is some information redundancy in the extracted details, a normalized mutual information is used to calculate the ratio of the two types of details and generate a dual-scale detail image through linear weighting. To ensure that the injected details align with the original panchromatic image, edge detection is performed using gradient operators on the enhanced multispectral image, and the resulting edge matrix serves as a constraint for detail injection. Additionally, the spectral preservation coefficient is calculated based on the spectral correlation between the original multispectral image and its intensity component to maintain the spectral relationship during the fusion process. Finally, the dual-scale detail image, edge detection matrix, and spectral preservation coefficient are multiplied to obtain the final detail injection amount. The third stage is image fusion, where the original upsampled multispectral image is pixel-wise added to the detail injection image from the second stage to generate the fused high-resolution multispectral image. To validate the effectiveness of the proposed method, experiments were conducted on four types of remote sensing datasets including IKONOS, QuickBird, WorldView4, and GF-2. The images used in the experiments contain various terrain elements such as vegetation, buildings, and water bodies to verify the requirements of different land features for spectral and spatial information. Comparing the results of the four sets of experiments, the proposed method shows color similarity to the original multispectral images in the fused true-color composite images, with clear edges and rich textures. In terms of objective quality evaluation, the experiments on IKONOS and WorldView4 datasets achieved the best or second-best results in D_λ, D_s and QNR, while in the remaining two indicators, although there was no significant advantage, the subjective visual quality was significantly better than the comparison methods. In conclusion, the proposed method, combining spectral and spatial scale detail injection, addresses the shortcomings of insufficient single-scale detail information extraction, better adapts to the characteristics of different land features, and improves the detail and accuracy of the fusion results.