Air strikes are among the main means of attack in modern warfare. To improve air defense capabilities and aid military decision-making, threat assessment models have been introduced. As the parameters of the kernel extreme learning machine (KELM) model need to be set individually, this study proposes a parameter learning strategy based on a multistrategy improved sparrow search algorithm (MISSA). First, a reasonable threat assessment model was established based on the capability and situation factors of air targets. Second, the sparrow search algorithm was improved in terms of population position initialization and position update strategy, incorporating tent chaos reverse learning, nonlinear inertia weights, a global search strategy, and adaptive t-distribution. The effectiveness of the MISSA strategy was verified using nine common benchmark functions. The results show that the proposed MISSA finds an effective balance between global and local searches. Moreover, when the MISSA is applied to solve the tuning problem of KELM, the values of mean absolute percentage error, mean square error, root mean square error, and mean absolute error for MISSA–KELM in the air target threat assessment problem are 2.013 × 10−2, 1.282 × 10−4, 1.132 × 10−2, and 8.316 × 10−3, respectively, all of which are higher than that of the other metaheuristic algorithms (e.g., ACWOA-KELM and HGWO-KELM). Therefore, the method proposed in this study can be used as a parameter-tuning tool for KELM, enabling KELM to perform better in practical applications.

Semantic segmentation of remote sensing images poses a formidable challenge within this domain. Our investigation commences with a pilot study aimed at scrutinizing the advantages and disadvantages of employing a Transformer architecture and a CNN architecture in remote sensing imagery (RSI). Our objective is to substantiate the indispensability of both local and global information for RSI analysis. In this research article, we harness the potential of the Transformer model to establish global contextual understanding while incorporating an additional convolution module for localized perception. Nonetheless, a direct fusion of these heterogeneous information sources often yields subpar outcomes. To address this limitation, we propose an innovative hierarchical fusion feature information module that this model can fuse Transformer and CNN features using an ensemble-to-set approach, thereby enhancing information compatibility. Our proposed model, named FURSformer, amalgamates the strengths of the Transformer architecture and CNN. The experimental results clearly demonstrate the effectiveness of this approach. Notably, our model achieved an outstanding accuracy of 90.78% mAccuracy on the DLRSD dataset.