Storage capacity of an optically formed spatial filter for character recognition

Area of Science:

• Optics
• Information Science
• Image Processing

Background:

• Optical spatial filtering is a proposed method for character recognition.
• This technique involves performing optical cross-correlations between unknown characters and stored masks.

Purpose of the Study:

• To derive estimates for the storage capacity of optical character recognition systems using spatial filtering.
• To evaluate the impact of photographic plate noise on system capacity.

Main Methods:

• Derivation of theoretical capacity estimates for both noiseless and noisy optical systems.
• Experimental noise measurements were conducted on Kodak 649F photographic plates.
• A numerical example was used to demonstrate the order of magnitude of the system capacity.

Main Results:

• A noiseless optical system demonstrated a theoretical capacity of several hundred thousand storable masks.
• The presence of photographic plate noise reduced the system's capacity by two orders of magnitude.
• Noise measurements provided empirical data for the noisy system estimate.

Conclusions:

• The storage capacity of optical spatial filtering systems for character recognition is significantly limited by photographic plate noise.
• The theoretical capacity is high in ideal conditions, but practical limitations due to noise must be considered for real-world applications.