Motion image coder using pre-filter to reduce quantization error

Claim 2

A motion image coding apparatus comprising: a first field/frame memory for storing an input source image in units of fields or frames in accordance with a coding mode; a pre-filter for adaptively restricting a band for characteristics of coded blocks with respect to the source image stored in said first field/frame memory, said pre-filter including a passband index, wherein said pre-filter classifies characteristics of the input source image into n classes according to the complexity of the image signal and calculates the passband index according to the classification, to thereby perform low-pass-filtering per macroblock of the input source image and outputting a current image signal; a subtractor for subtracting a motion-compensated previous image signal from the current image signal output from said pre-filter and outputting a difference image signal; a discrete cosine transform (DCT) circuit for performing a DCT function with respect to the difference image signal output from said subtractor; a quantizer for quantizing the output signal of said DCT circuit into a predetermined quantization step size; a variable-length coder for variable-length-coding the output signal of said quantizer; a multiplexer for multiplexing an output signal of said variable-length coder, quantization parameters, and motion vectors, and outputting multiplexed data; a buffer for temporarily storing data output from said multiplexer and transmitting the stored data to a receiver side at a constant rate; a bit-rate controller for determining a base quantization step size by the data fullness of said buffer; an activity calculator for calculating the image activity from the base quantization step size output from said bit-rate controller and the passband index of said pre-filter, thereby determining a final quantization step size based on the image activity for controlling said quantizer; an inverse quantizer for restoring the signal quantized in said quantizer into the original signal before the quantization; an inverse DCT circuit for restoring the signal inverse-quantized in said inverse quantizer into the original signal before DCT transformation by said DCT circuit and outputting an image signal; an adder for adding the restored image signal output from said inverse DCT circuit to the motion-compensated previous image signal to produce a motion-compensated restored image signal; a second field/frame memory for storing the motion-compensated restored image signal output from said adder in units of fields or frames in accordance with a coding mode; a motion estimator for generating motion vectors for constituting a current image depending on a previous image output from said pre-filter; and an adaptive predictor for compensating motion position of the motion-compensated restored image signal stored in said second field/frame memory with the motion vectors output from said motion estimator.

Claim 3

A motion image coding apparatus comprising: a first field/frame memory for storing a source image in units of fields or frames in accordance with a coding mode; a pre-filter for adaptively restricting a band for characteristics of coded blocks with respect to the source image stored in said first field/frame memory, said pre-filter including a passband index and outputting a current image signal; a subtractor for subtracting a motion-compensated previous image signal from the current image signal output from said pre-filter and outputting a difference image signal; a discrete cosine transform (DCT) circuit for performing a DCT function with respect to the difference image signal output from said subtractor; a quantizer for quantizing the output signal of said DCT circuit into a predetermined quantization step size; a variable-length coder for variable-length-coding the output signal of said quantizer; a multiplexer for multiplexing an output signal of said variable-length coder, quantization parameters, and motion vectors, and outputting multiplexed data; a buffer for temporarily storing data output from said multiplexer and transmitting the stored data to a receiver side at a constant rate; a bit-rate controller for determining a base quantization step size by the data fullness of said buffer; an activity calculator for calculating the image activity from the base quantization step size output from said bit-rate controller and the passband index of said pre-filter, thereby determining a final quantization step size based on the activity for controlling said quantizer, wherein said activity calculator calculates the activity by multiplying a base quantization parameter, an activity parameter obtained by a forward activity calculation for an evaluation of macroblock complexity for performing a coding, and a band-restricted compensation parameter obtained using the passband index; an inverse quantizer for restoring the signal quantized in said quantizer into the original signal before the quantization; an inverse DCT circuit for restoring the signal inverse-quantized in said inverse quantizer into the original signal before DCT transformation by said DCT circuit and outputting a restored image signal; an adder for adding the restored image signal output from said inverse DCT circuit to the motion-compensated previous image signal to produce a motion-compensated restored image signal; a second field/frame memory for storing the motion-compensated restored image signal output from said adder in units of fields or frames in accordance with a coding mode; a motion estimator for generating motion vectors for constituting a current image depending on a previous image output from said pre-filter; and an adaptive predictor for compensating motion position of the motion-compensated restored image signal stored in said second field/frame memory with the motion vectors output from said motion estimator.

Claim 4

A motion image coding apparatus comprising: a first field/frame memory for storing a source image in units of fields or frames in accordance with a coding mode; a pre-filter for adaptively restricting a band for characteristics of coded blocks with respect to the source image stored in said first field/frame memory, said pre-filter including a passband index and outputting a current image signal; a subtractor for subtracting a motion-compensated previous image signal from the current image signal output from said pre-filter and outputting a difference image signal; a discrete cosine transform (DCT) circuit for performing a DCT function with respect to the difference image signal output from said subtractor; a quantizer for quantizing the output signal of said DCT circuit into a predetermined quantization step size; a variable-length coder for variable-length-coding the output signal of said quantizer; a multiplexer for multiplexing an output signal of said variable-length coder, quantization parameters, and motion vectors, and outputting multiplexed data; a buffer for temporarily storing data output from said multiplexer and transmitting the stored data to a receiver side at a constant rate; a bit-rate controller for determining a base quantization step size by the data fullness of said buffer; an activity calculator for calculating the image activity from the base quantization step size output from said bit-rate controller and the passband index of said pre-filter, thereby determining a final quantization step size based on the activity for controlling said quantizer, wherein said activity calculator calculates the activity by multiplying a base quantization parameter, an activity parameter obtained by a forward activity calculation for an evaluation of macroblock complexity for performing a coding, and a band-restricted compensation parameter obtained using the passband index; an inverse quantizer for restoring the signal quantized in said quantizer into the original signal before the quantization; an inverse DCT circuit for restoring the signal inverse-quantized in said inverse quantizer into the original signal before DCT transformation by said DCT circuit and outputting a restored image signal; an adder for adding the restored image signal output from said inverse DCT circuit to the motion-compensated previous image signal to produce a motion-compensated restored image signal; a second field/frame memory for storing the motion-compensated restored image signal output from said adder in units of fields or frames in accordance with a coding mode; a motion estimator for generating motion vectors for constituting a current image depending on a previous image output from said pre-filter; and an adaptive predictor for compensating motion position of the motion-compensated restored image signal stored in said second field/frame memory with the motion vectors output from said motion estimator.