Fixed-weight coding method for uniformity processing in holographic data storage

Holographic data storage technology leverages the principles of laser interference to record two-dimensional data pages in a three-dimensional manner within a photosensitive medium. It boasts features such as high data storage capacity, long life, and fast data transmission rates. However, the non-uniform diffraction efficiency of the recording medium leads to distortion in the energy distribution of the reconstructed images. This issue impairs the training performance of the deep learning decoding network model, while also leading to a high bit error ratio when the traditional threshold method is adopted for decoding. To address this issue, this paper proposes a fixed-weight multi-level data page encoding method to ensure that data pages achieve a uniform energy distribution during the encoding phase. On this basis, an adaptive homomorphic filtering method is further introduced. This method is based on the uniform energy characteristics of fixed-weight data pages and reduces the variance of the energy distribution of reconstructed images. Subsequently, the filtered dataset is used to train a complex amplitude decoding convolutional neural network, while the known encoding information is utilized during the decoding process to directly mitigate the impact of uneven energy distribution. Ultimately, this approach reduces the bit error ratio of complex-amplitude data by over 60%. The proposed data page encoding approach and the adaptive homomorphic filtering method provide robust support for the further development of holographic storage and hold potential for cross-domain applications.