We suggest a new method for data detection based on direct data identification and compare this new method with three commonly used methods based on decimation using linear transformations. To compare these methods, the complete procedure from data page creation, sending data through an optical channel, and data detection is simulated numerically. This is performed by varying the noise level, sampling factor, and spatial filter. The proposed method requires a higher computational complexity, but for a broad range of realistic system parameters, it enables lower symbol error ratios.