This 1970 lost film demonstrates the U.S. Postal Service's Optical Character Recognition (OCR) machines, which allow mail to be sorted automatically. This was one of the earliest mass scale deployments of this technology. https://t.co/ch5GswTaag
Video Transcript AI Summary
The video showcases the optical character reading machines used by the post office department. These machines can read 42,000 addresses per hour and sort mail based on different sort schemes. The process involves clerks feeding letter-sized mail into the input hopper, where the machines scan and read the addresses. The machines use a modulated beam of light to recognize individual characters by detecting distinctive features. The recognized characters are stored in a magnetic drum memory for comparison with incoming or outgoing addresses. Once the addresses are matched, the mail is directed to the appropriate sorting receptacle. This automated process is mainly used for business mail and helps expedite the processing and dispatching of mail.
Speaker 0: The optical character reading machines in use by the post office department are among the more complex electronic devices. They are expected to read at a speed of 42,000 addresses an hour. And sort mail to as many separations as required by the sort schemes in a particular post office. Separations to large cities, to states, and to sectional centers. This film We'll show you how this remarkable optical character reading machine performs its functions today in several of our large city post offices.
This is a typical installation of an optical character reader in a major postal facility. Clerks at 2 input stations feed letter size mail into the input hopper. The address side Enter the letters are all faced in the same direction. Rate of speed at each input station is 21,000 letters Per hour, 350 per minute. To make most efficient use of the electronic reading section and memory system, One scanner is reading an address on an envelope during a time span of 1 12th of a second.
While another letter already read is being transported past the other scanner. This permits a machine read rate of 42,000 letters per hour and sorting to the required number of separations at the rear the machine at the same speed. Here's how reading takes place. The optical character reader does 3 things. One, it finds the bottom line Or the 2nd line from the bottom of an address depending on whether an outgoing or an incoming sort is required.
Two, it measures the height of the 1st character of the chosen line. In this particular instance, the height of the upper case f. Three, it reads the line. Here is how the reader does these 3 things. A modulated beam of light generated by the phosphor coating on the face of a cathode ray tube operates through an expanding optical system.
This optical system projects a 2 and a half inch by 8 inch coarse raster scan at the address side of the mail piece. When this scan encounters what appears to be an address block, it continues scanning until it passes the leftmost character of the bottom line. This operation called line find takes 5 thousandths of a second. The raster lines then change to a fine scan in order to measure the height of the first character in the bottom line. In this instance, the upper case f.
The reader now knows the height of the first character in the line and sets its read scan to this dimension and proceeds to scan each character in turn at the rate of 1,000 characters per second. Feature detection is the method used by the optical reader to actually recognize individual characters. By feature detection, we mean recognizing distinctive features which make up a character. Notice that these characters all have straight vertical lines. Some have curves open to the left.
2 have lines slanting downward to the right. And 1 has a line slanted upward to the right. These are called features. They are used to make up many different characters. For instance, a vertical line and a curve open to the left when in the proper relationship form the character p.
A vertical line and 2 curves open to the left form the character b. For the optical reader, The recognition of characters has been reduced to the task of recognizing features. In order to detect all the fonts, both upper and lower case, And numerics, the optical character reader is programmed with some 300 of these features. A vertical line, a horizontal line at about the center, and a horizontal line at the top. This combination yields recognition of the character F.
To accomplish this feature recognition during the ReED scan, the reader samples 576 points in the character space To determine whether there is black or white information at each of these points. Starting at the lower left corner And moving up each vertical column, the reader obtains an information sequence such as black, black, black, white, white, white, and so on. In the 2nd column, the information sequence would show black, black, black, black, white, white, white, and so on. This continues until the reader has obtained a sequence of information about the 576 points. Now the reader has determined that the character consists of a vertical line on the left, a horizontal line across the center, and a horizontal line across the top.
That is, it has recognized the salient features of this character. Here on a readout test device, We see several characters recognized and formed electronically in the recognition unit. Now the reader stores each character until it has duplicated and assembled the entire line. The line that it has read. This cabinet contains a 10,000,000 bit magnetic drum memory.
Forty thousand addresses are programmed. Each of the tracks contains about 18,000 bits of information. 300 bits per inch of drum. The drum revolves at a rate of 30 times per second. This speed enables the memory drum to compare the address found in the read operation with all the incoming or outgoing addresses it has in storage.
Within a total line find, height find, and read period of 83 milliseconds, and a compare period of another 80 milliseconds, The address line has been read, and the mail piece is directed to the proper sorting receptacle. The comparison unit activates wheel setters to set the code wheels of each pocket for a particular receptacle. The door of the carrier pocket opens when it reaches a receptacle with a matching code reel combination. The letter is now ready for dispatch to a city and state. If the 2nd line, incoming mail, had been read, The letter would be sorted to a specific city carrier or carrier station.
In our post offices today, Mail which is automatically read and sorted is mostly business mail, specifically tagged for OCR, optical character reader sorting. It moves directly from the incoming platform to the OCR machines and is processed very rapidly and economically. The post office department is leading the search for better ways to process and dispatch mail promptly, Accurately and in the shortest time possible.
