Parallel processing of digital signals in a single arithmetic/logic unit

A method of processing signals in an arithmetic/logic unit having a register including a plurality of segments, each segment accommodating predetermined numbers of bits, and in which carry bits are propagated by operations in the arithmetic/logic unit from a lower order segment of the register to a higher order segment of the register, the method comprising the steps of:

• receiving a digital signal formed of a sequence of digital samples representative of numerical values, wherein the digital samples form sets in which at least one anticipated processing step is performed identically upon each digital sample in each set and wherein each set includes a digital sample represented by a number of bits corresponding to one of the plurality of segments;
• loading into contiguous adjacent segments of a single register, without unused bits therebetween, a set of first operands taken from a group including a first set of the digital samples and a set of digital samples arranged to have correspondence with the segments of the single register, a second set of operands formed by a remainder of the group;
• performing an arithmetic/logic unit function that causes carry bits to be propagated, the function performed simultaneously between all the digital samples of the first set of the operands and the second set of operands, forming a new set of digital samples in the register;
• and
• providing as a processed digital signal, a new sequence of the new set of digital samples.

A method of processing digital image signals, comprising the steps of:

• representing the digital image signal as a sequence of pixel signals;
• loading into adjacent segments in a single register a first plurality of pixel signals;
• subtracting from the first plurality of pixel signals a second plurality of pixel signals, thereby producing in the single register a plurality of difference signals;
• propagating an error introduced by one of the plurality of difference signals to another of the plurality of difference signals;
• comparing the difference signals with a predetermined range;
• adding to the difference signals a third plurality of pixel signals, the third plurality of pixel signals and the second plurality of pixel signals having values related by a first constant factor, thereby producing a fourth plurality of pixel signals;
• selecting as an output plurality of pixel signals the first plurality of pixel signals scaled by a second constant factor when any of the plurality of difference signals lies outside the predetermined range and selecting as the output plurality of pixel signals the fourth plurality of pixel signals when none of the plurality of difference signals lies outside the predetermined range;
• delaying the output plurality of pixel signals by one frame, thereby producing the third plurality of pixel signals;
• and
• assembling the output plurality of pixel signals into a sequence forming a new digital image signal.

A method of processing image signals, comprising the steps of:

• receiving an image signal;
• representing the image signal as a digital image signal formed of a sequence of digital samples representative of numerical values;
• categorizing the digital samples into sets, each set including a plurality of digital samples;
• loading into adjacent segments of a single register a first set of digital samples;
• adding to the first set of digital samples in the single register a second set of digital samples, forming a new set of digital samples in said adjacent segments;
• propagating an error introduced by one of the adjacent segments to another of the adjacent segments;
• multiplying the new set of digital samples in said adjacent segments by a constant;
• and
• adding to the multiplied new set of digital samples in the register a third set of digital samples;
• propagating an error introduced by one of the adjacent segments to another of the adjacent segments.

A method of processing image signals, comprising the steps of:

• receiving an image signal representative of an image having at least two orthogonal dimensions;
• representing the image signal as a digital image signal formed of a sequence of digital samples representative of numerical values;
• categorizing the digital samples into sets, each set including a plurality of digital samples taken along a first dimension of the two orthogonal dimensions;
• loading into adjacent segments of a single register a first set of digital samples;
• adding to the first set of digital samples in the single register a second set of digital samples, forming a set of new digital samples in said adjacent segments;
• forming a new sequence of the new digital samples;
• categorizing the new digital samples into sets of new digital samples taken along a second dimension of the two orthogonal dimensions;
• loading into the adjacent segments of the single register a first set of new digital samples;
• and
• adding to the first set of new digital samples in the single register a second set of new digital samples.