In order to improve the image quality of the aerial optoelectronic sensor over a wide range of temperature changes, high thermal adaptability of the primary mirror as the critical components is considered. Integrated optomechanical analysis and optimization for mounting primary mirrors are carried out. The mirror surface shape error caused by uniform temperature decrease was treated as the objective function, and the fundamental frequency of the mirror assembly and the surface shape error caused by gravity parallel or vertical to the optical axis are taken as the constraints. A detailed size optimization is conducted to optimize its dimension parameters. Sensitivities of the optical system performance with respect to the size parameters are further evaluated. The configuration of the primary mirror and the flexure are obtained. The simulated optimization results show that the size parameters differently affect the optical performance and which factors are the key. The mirror surface shape error under 30 °C uniform temperature decrease effectively decreased from 26.5 nm to 11.6 nm, despite the weight of the primary mirror assembly increases by 0.3 kg. Compared to the initial design, the value of the system’s modulation transfer function (0° field angle) is improved from 0.15 to 0.21. Namely, the optical performance of the camera under thermal load has been enhanced and thermal adaptability of the primary mirror has been obviously reinforced after optimization. Based on the optimized results, a prototype of the primary mirror assembly is manufactured and assembled. A ground thermal test was conducted to verify difference in imaging quality at room and low temperature, respectively. The image quality of the camera meets the requirements of the index despite degrading.

Most single-object trackers currently employ either a convolutional neural network (CNN) or a vision transformer as the backbone for object tracking. In CNNs, convolutional operations excel at extracting local features but struggle to capture global representations. On the other hand, vision transformers utilize cascaded self-attention modules to capture long-range feature dependencies but may overlook local feature details. To address these limitations, we propose a target-tracking algorithm called CVTrack, which leverages a parallel dual-branch backbone network combining CNN and Transformer for feature extraction and fusion. Firstly, CVTrack utilizes a parallel dual-branch feature extraction network with CNN and transformer branches to extract local and global features from the input image. Through bidirectional information interaction channels, the local features from the CNN branch and the global features from the transformer branch are able to interact and fuse information effectively. Secondly, deep cross-correlation operations and transformer-based methods are employed to fuse the template and search region features, enabling comprehensive interaction between them. Subsequently, the fused features are fed into the prediction module to accomplish the object-tracking task. Our tracker achieves state-of-the-art performance on five benchmark datasets while maintaining real-time execution speed. Finally, we conduct ablation studies to demonstrate the efficacy of each module in the parallel dual-branch feature extraction backbone network.