Skip to Main Content
Statistical studies of television signals indicated a high degree of correlation between successive elements, lines and frames. A continuous test run indicated that the normalized detail content of actual broadcast signals run lower than 5 percent. The fact that most transmitted television scenes thus measured contained very little detail compared to some artificial subjects indicated that the observer's preference has influenced the subject to be transmitted. In order to verify this assumption, some tests were devised to measure the perception speed of observers. These tests included certain reading and character recognition tests and finally a test consisting cf object recognition in precisely measured periods was devised. In order to evaluate the perception rate in bits, a relationship between the gestalt concept and the binary choices was sought. Assuming a limited number of picturable nouns to correspond to the gestalt experience of the observer, each object recognized was assigned a value of ten bits (corresponding to 1024 picturable nouns). Several series of these tests indicated that the visual perception speed of a normal observer is between 30 to 50 bits per second, that this value holds for periods of one-tenth to two seconds, and that the first thing observed is the center of the picture. These values are similar to the values obtained by Licklider, Pierce and others for reading speeds.