Bit length converting circuit for floating-point numbers, and vibration correction control circuit using the bit length converting circuit
The bit length reducing unit reduces the bit length of an operand expressed by a floating-point number and outputs the operand to a computing unit. The bit length increasing unit increases the bit length of the result value of operation represented by the floating-point number inputted from the computing unit and restores the original bit length. The bit length reducing unit discards a preset number of higher-order bits of the exponent part of the floating-point number and at the same time adds a positive or negative offset value to the exponent part with the higher-order bits discarded, according to an application to be executed by the computing unit. The bit length increasing unit restores the bits discarded by the bit length reducing unit 11 from the exponent part of the floating-point number and at the same time subtracts an offset value from the exponent part.
1. A vibration correction control circuit for correcting an optical axis of an image pickup apparatus in response to a vibration thereof, the circuit comprising:
a vibration-component signal generating unit configured to generate a vibration-component signal indicating an amount of movement of an image pickup apparatus according to an output signal of a vibration detecting element for detecting a vibration of the image pickup apparatus;
a driver signal generating unit configured to generate a drive signal used to control a driver element to correct a position of a lens, based on an output signal of a position detecting element for detecting the position of the lens to be driven and the vibration-component signal; and
a bit length converting circuit for a floating-point number configured to convert the bit length of the output signal of the vibration detecting element expressed by the floating-point number, the bit length converting circuit comprising:
a bit length reducing unit configured to reduce a bit length of an operand expressed by the floating-point number and configured to output the operand to a computing unit; and
a bit length increasing unit configured to increase the bit length of a result value of operation represented by the floating-point number inputted from the computing unit and configured to restore an original bit length, wherein said bit length reducing unit discards a predetermined number of higher-order bits of the exponent part of the floating-point number and adds a positive or negative offset value to the exponent part with the higher-order bits discarded, according to an application to be executed by the computing unit, and wherein said bit length increasing unit restores a bit discarded by the bit length reducing unit from the exponent part of the floating-point number and subtracts the offset value from the exponent part thereof, and wherein said vibration-component signal generating unit includes: a high-pass filter configured to filter the output signal of the vibration detecting element, whose bit length has been converted by said bit length converting circuit, with a preset cutoff frequency; and an integration circuit configured to generate the vibration-component signal by integrating an output signal of the high-pass filter, wherein the high-pass filter and the integrating circuit comprises digital filters.
2. A bit length converting circuit for floating-point numbers according to claim 1 , wherein said bit length reducing unit discards a predetermined number of lower-order bits of the fraction part of the floating-point number by performing a rounding processing on the predetermined number of lower order bits thereof, and wherein said bit length increasing unit places zeros in bit positions, of the fraction part of the floating-point number, which is discarded by said bit length reducing unit and restores the bit length of the fraction part thereof to the original bit length.
3. A vibration correction control circuit according to claim 1 , wherein said bit length reducing unit discards a predetermined number of higher-order bits of the exponent part of the floating-point number and adds a negative offset value to the exponent part with the higher-order bits discarded, according to an arithmetic operation processing executed by the high-pass filter and the integration circuit.
4. The bit length converting circuit of claim 1 , wherein the bit length increasing unit comprises:
a comparing unit;
a subtracting unit coupled to the comparing unit; and
a bit adding unit coupled to the comparing unit and the subtracting unit.
5. A vibration correction control circuit for correcting an optical axis of an image pickup apparatus in response to a vibration thereof, the circuit comprising:
a vibration-component signal generating unit configured to generate a vibration-component signal indicating an amount of movement of an image pickup apparatus according to an output signal of a vibration detecting element for detecting a vibration of the image pickup apparatus;
a driver signal generating unit configured to generate a drive signal used to control a driver element to correct a position of image pickup devices, based on an output signal of a position detecting element for detecting the position of the image pickup devices to be driven and the vibration-component signal; and
a bit length converting circuit for floating-point numbers configured to convert the bit length of the output signal of the vibration detecting element expressed by the floating-point numbers, the bit length converting circuit comprising:
a bit length reducing unit configured to reduce a bit length of an operand expressed by a floating-point number of the floating point numbers and configured to output the operand to a computing unit; and
a bit length increasing unit configured to increase the bit length of a result value of operation represented by the floating-point number inputted from the computing unit and configured to restore an original bit length, wherein said bit length reducing unit discards a predetermined number of higher-order bits of the exponent part of the floating-point number and adds a positive or negative offset value to the exponent part with the higher-order bits discarded, according to an application to be executed by the computing unit, and wherein said bit length increasing unit restores a bit discarded by the bit length reducing unit from the exponent part of the floating-point number and subtracts the offset value from the exponent part thereof, and wherein, said vibration-component signal generating unit including: a high-pass filter configured to filter the output signal of the vibration detecting element, whose bit length has been converted by said bit length converting circuit, with a preset cutoff frequency; and an integration circuit configured to generate the vibration-component signal by integrating an output signal of the high-pass filter, wherein the high-pass filter and the integrating circuit comprises digital filters.
6. A vibration correction control circuit according to claim 5 , wherein said bit length reducing unit discards a predetermined number of higher-order bits of the exponent part of the floating-point number and adds a negative offset value to the exponent part with the higher-order bits discarded, according to an arithmetic operation processing executed by the high-pass filter and the integration circuit.
7. The bit length converting circuit of claim 5 , wherein the bit length increasing unit comprises:
a comparing unit;
a subtracting unit coupled to the comparing unit; and
a bit adding unit coupled to the comparing unit and the subtracting unit.
8. A bit length converting circuit for floating-point numbers according to claim 5 , wherein said bit length reducing unit discards a predetermined number of lower-order bits of the fraction part of the floating-point number by performing a rounding processing on the predetermined number of lower order bits thereof, and wherein said bit length increasing unit places zeros in bit positions, of the fraction part of the floating-point number, which is discarded by said bit length reducing unit and restores the bit length of the fraction part thereof to the original bit length.