Abstract: A theoretically optimum motor drive time period or drive pulse width is computed as a function of the number of steps from the present motor shaft step to the desired command step. A compensation time period is computed as a function of a previous motor drive time period and combined with the computed drive time period to produce a present drive time period for energization of the motor (12) for movement of the motor shaft (14) from the present step to the next step.

Abstract: A theoretically optimum motor drive time period or drive pulse width is computed as a function of the number of steps from the present motor shaft position to the desired command position and the present motor shaft rotational speed. A compensation factor is computed as a function of the present motor shaft speed and the torque-speed characteristic of the motor (12) and multiplied by the computed drive time period to produce a present drive time period for energization of the motor (12) for rotation of the motor shaft (14) from the present position to the next step position. After the motor shaft (14) has been rotated to within one step of the command position, sensor means (18') sense displacement of the motor shaft (14) in excess of a predetermined small angular distance from the command position and cause the motor (12) to be energized to rotate the motor shaft (14) toward the command position.

Abstract: A character is scanned in a plurality of parallel scan lines and the starting and ending positions of the character in the respective scan lines are detected. The portions of the character between the starting and ending positions of the respective scan lines are quantized and classified into a code. The starting and ending positions and classification codes are compared with those of standard character patterns and the character is recognized as the standard character with the closest correlation. This operation corresponds to tracing the right and left outlines of the character, and further classifying the interior of the character. Means are provided for normalizing the horizontal size, position and slant of the character prior to correlation. An improved quadratic correlation function is disclosed which selectively utilizes root mean square values of positive and negative projection components of a characteristic vector of a covariance matrix of a standard character pattern.

Abstract: Line portions of the character pattern in a character recognition device which have a thickness less than a selected standard value are thickened by a thickening circuit to the selected value. The line thickness of the character is then uniformly thinned by a thinning circuit and the thickening process is repeated. A correction circuit may be provided to prevent the continuity of the character from being broken in the thinning circuit.

Abstract: The character pattern from the scanner is stored in its original form in an auxiliary memory and in compressed form in a main memory. The slant and the horizontal and vertical second-order moments of the character pattern are sensed in the main memory, and normalized as the character pattern is shifted from the auxiliary memory to the main memory.

Abstract: In a pattern recognition system of the type in which an input pattern is read, and stored in a two-dimensional memory, and after the features of the input pattern are extracted, they are compared with the standard patterns in storage in such a way that the degrees of similarity of the input pattern to the standard patterns are obtained so as to establish one-to-one correspondence between the input pattern and one of the standard patterns in storage that has the maximum degree of similarity, a part of the input pattern is enlarged when the degrees of similarity of the input pattern to two of the standard patterns are in excess of a predetermined level, so that the features of the enlarged part of the input pattern information may be extracted so as to establish one-to-one correspondence between the input pattern and one of said two standard patterns. Additionally, a reject signal is provided when the input pattern has a greater-than-predetermined similarity to three or more standard patterns.