ABSTRACT
One of many factors influencing the accuracy of photogrammetric outputs is the quality of image data. This is especially important in high precision measurements, where we meet with subpixel measurement accuracy and therefore the circular or coded targets are used. The accuracy of Least Square Matching (LSM) method depends on similarity between template and search image and target diameter and can reach theoretical value of 0.005 pixel. Similarity of images is strictly given by perspective, projective and optical distortions and radiometric differences. We know that image compression changes radiometric properties of some pixels which could have an impact on measurement of image coordinates by the LSM method. In this article we empirically determine the influence of various image compression on the measurement accuracy of points in the image plane and on resulting 3D coordinates in reference system. The influence of different size and shape of circle targets in combination with image compression is also tested. The coded targets on a special testing frame were evaluated using the multi-image convergent photogrammetry. The RAW images were converted into various image formats with different compression and the influence of this compression on point measurement accuracy was estimated.
