Based on micro-indentation mechanics and kinematics of grinding processes, theoretical formulas are deduced to calculate surface roughness (SR) and subsurface damage (SSD) depth. The SRs and SSD depths of a series of fused silica samples, which are prepared under different grinding parameters, are measured. By experimental and theoretical analysis, the relationship between SR and SSD depth is discussed. The effect of grinding parameters on SR and SSD depth is investigated quantitatively. The results show that SR and SSD depth decrease with the increase of wheel speed or the decrease of feed speed as well as cutting depth. The interaction effect between wheel speed and feed speed should be emphasized greatly. Furthermore, a relationship model between SSD depth and grinding parameters is established, which could be employed to evaluate SSD depth efficiently.

Infrared small target (IST) detection plays a critical role in both civilian and military applications. However, in infrared remote sensing images, targets typically appear small, lack detailed textural information, low energy and resolution. Previous object detection methods encounter difficulties in accurately identifying targets and avoiding mistakenly identifying interference as targets when detecting IST in complex background environments. To address these issues, we propose STASPPNet, an IST detection network based on a Swin Transformer and multi-scale atrous spatial pyramid pooling, which can detect infrared small targets (ISTs) in complex backgrounds. A Swin-Conv backbone is designed to enhance feature extraction for ISTs. Additionally, we create the multi-scale atrous spatial pyramid pooling module, leveraging atrous convolutions and max pooling layers to improve feature representation, leading to more accurate detection. Moreover, we develop the multi-scale feature detection head, enabling the detection of targets in infrared images with as few as a dozen pixels and enhancing the performance for weak infrared targets. Extensive experiments illustrate that STASPPNet achieves an average accuracy of 96.9% for mAP50 and 49.7% for mAP50-95 on the NUAA-SIRST dataset and outperforms other state-of-the-art networks. The ablation experiments prove the effectiveness of the proposed module and network. The source code of the STASPPNet will be available at https://github.com/hwuscut/STASPPNet.

Chemical etching experiments are carried out to expose and change the morphology of multiple scratches and single scratches in a set of K9 glasses, during which the transmission of some glasses is measured. The results show that brittle scratches have an influence on the transmission, and the transmission decreases at first and then increases obviously with the increase of etching time for the glass with a high density of cracks. The scattered field distribution around trailing indent crack(s) and the transmission of the glass with a scratch etched deeply (adjacent cracks have coalesced with each other) are simulated under different parameters. Using finite-difference time-domain (FDTD) algorithm, the light intensification effect around two single-evolving scratches and two adjacent cracks with different parameters are simulated, respectively. The results indicate that the position with the maximum light intensity changes for the glass with a scratch and its light intensity enhancement factors (LIEFs) will generally decrease as the etching process progresses. Compared with the morphology evolution of a scratch during the etching process, the changing mechanisms of transmission and LIEFs are revealed. This work will contribute to the improvement of service performance of optical glass by chemical etching process.