Method of parallelly filtering input data words to obtain final output data words containing packed half-pel pixels
View Patent ↗The present invention is to provide a parallel filtering method, which is implemented to an interpolation filter and comprises the steps of separating coefficients of the interpolation filter into two sets comprising the positive and negative coefficients respectively for parallelly filtering a plurality of input data pixels packed into data words inputted to the interpolation filter concurrently to obtain a first result data word, and clipping and shifting the first result data word to obtain a final output data word containing packed half-pel pixels for parallelly and efficiently filtering data stream of video without increasing the complexity, cost, size and power consumption of circuitry of an electronic video apparatus.
1. A parallel filtering method, which is implemented to an interpolation filter embedded in an electronic video apparatus on executable memory, comprising the steps:
performing on a first circuit the steps:
separating coefficients of the interpolation filter into two sets comprising positive and negative coefficients respectively for parallelly filtering a plurality of input data pixels packed into data words inputted to the interpolation filter concurrently;
multiplying the data words with absolute values of the positive and negative coefficients respectively;
accumulating all the data words multiplied by the absolute values of the positive coefficients to produce a first intermediate data word, and adding a first constant data word denoting the sum of rounding offset and 0x80008000 to the first intermediate data word to obtain a second intermediate data word;
accumulating all the data words multiplied by the absolute values of the negative coefficients to obtain a third intermediate data word;
subtracting the third intermediate data word from the second intermediate data word to obtain a first result data word; and
performing on a second circuit the steps:
performing a first AND logical operation between the first result data word and a second constant data word to obtain a fourth intermediate data word;
determining whether or not the fourth intermediate data word is equal to 0x80008000;
when it is determined that the fourth intermediate data word does not equal 0x80008000, performing a second AND logical operation between the first result data word and 0x80008000 to obtain a fifth intermediate data word;
subtracting the fifth intermediate data word right shifting by a first predetermined bit position from the fifth intermediate data word to obtain a first mask data word;
performing a third AND logical operation between the first mask data word and the first result data word to obtain a second result data word;
performing a fourth AND logical operation between the second result data word and a third constant data word to obtain a sixth intermediate data word;
subtracting the sixth intermediate data word right shifting by a second predetermined bit position from the sixth intermediate data word to obtain a second mask data word; and
performing an OR logical operation between the second mask data word and the second result data word and then right shifting the result thereof by a third predetermined bit position to obtain a final output data word outputted through the electronic video apparatus.