针对传统眼轴长度测量需要分段检测且测量误差较大的问题, 设计了一种具有大探测范围的扫频光学相干层析系统, 实现了对眼轴长度的单次完整测量.提出了自适应峰值点提取与自适应误差校正相结合的算法,实现了大深度干涉信号的重构; 利用CPU-GPU协同加速技术实现了系统的实时测量, 解决了大范围探测数据量大、处理速度慢的问题.对光学眼模型进行实验, 结果表明：该系统对眼轴长度的测量误差为0.01 mm, 优于传统分段测量系统, 系统单次测量时间为0.10 s, 满足实时测量要求.

To solve the problem that the traditional measurement method of axial length is segmented detection and the measurement error is large, a swept-source optical coherence tomography system with large detection range and realized one-time measurement of axial length was designed. In order to reconstruct the large depth interference signal, an algorithm combining adaptive peak point extraction and adaptive error correction was proposed. The CPU-GPU acceleration technology was used to realize the real-time measurement and solve the problem of great data and slow speed of wide imaging range. Optical eye model experiments shows that the measurement error of axial length is 0.01 mm, which is superior to the traditional segment measurement system. The single measurement time is 0.10 seconds, which meets the real-time measurement requirements.