Optical transmission of barcodes using handheld cameras offers a practical solution for transmitting small amounts of data without device pairing. In various application scenarios, 2D matrix codes such as QR Code and Data Matrix have become ubiquitous for broadcasting short information such as web links. In recent years, both the increasing demand for higher data throughput and the rapid development of on-board cameras in smartphones have driven research into high-capacity systems. While the reduction of 2D symbol size can quadratically increase the spatial data capacity, accurate localization of small symbols, each consisting of only a few pixels, is extremely difficult due to the non-idealities of the camera-based optical channel. As a result, the data throughput achieved by state-of-the-art systems employing pattern-based methods is still far from optimal. In this work, we propose, for the first time to the best of our knowledge, a novel patternless approach to symbol position recovery for high-density matrix codes. We adapt the even-law synchronizer to estimate the symbol positions in each spatial dimension separately, and compensate for the impact of lens distortion by a successive curve fitting process. Simulation and experimental results show that our method significantly outperforms pattern-based benchmark methods by providing excellent accuracy and reliability.