This paper presents a new approach for damage detection in Structural Health Monitoring (SHM) Systems, which is based on Optical Scanning and Support Vector Machine (SVM) models. Optical Scanning Systems provide position measurements for SHM task by a novel method based on automatic geodetic measurements. Precise measurement of plane spatial angles are performed in the optical energy signal centre by the optical signal function geometric centroid calculation, however these scanners usually have non-linear variations in their measurement, and normally these variations depend on the position of the light emitter on the structure under monitoring in relation to the scanner. In this paper, SVM Regression is proposed as a machine learning technique to predict measurement errors and to adjust this non-linear variation for measurement accuracy enhancement.