Method of hydrid digital coding distinguishing motion pictures from still pictures

Claim 1

A method of hybrid digital coding of video signals, concerning both moving and still pictures, and organized into a sequence of frames which occur at a predetermined frequency and are formed by a sequence of lines each comprising a predetermined number of pixels, wherein: each frame is subdivided into pixel groups defining a plurality of hierarchical levels; for each frame there are coded, after quantization by a quantization law chosen for each group belonging to a first hierarchical level, coefficients of a two-dimensional transform applied either to a current frame or to the meaningful differences between the current frame and a predicted frame which has been possibly submitted to a motion compensation; and a coded signal is generated which comprises, among other information, information relevant to the coding type and to the quantization law employed, and which is asynchronously transmitted at variable bit rate, the quantization law adopted for a pixel group of the first hierarchical level using a quantization step determined by a coding control device (CCD) and varying depending on transmission bit availability, wherein, for each frame, the pixel groups are counted which belong to a second hierarchical level and which have been submitted to a predetermined coding type for a first number of previous consecutive frames, to recognize whether the frame belongs to a still or to a moving picture, and upon recognition of a still picture, for each group of pixels of such a picture belonging to said first level the quantization step is made to converge towards a predetermined minimum value (p1), which is the same for all groups, by the following operations: a) memorizing the quantization steps used for that group in the different frames of the still picture; b) detecting and memorizing a real quantization step (RQS) which is the minimum quantization step used till the current frame; c) comparing the real quantization step (R) and the quantization step determined for that group in the current frame by the coding control device (CCD); d) coding the transform coefficients relevant to that group with a quantization step (CQS) equal to that determined by the coding control device (CCD), if the latter step is lower than the real quantization step (RQS) or, if the quantization step determined by the coding control device is greater than or equal to the real quantization step (RQS), coding of the coefficients by a particular quantization step obtained by increasing by a first predetermined quantity (p2) the real quantization step, in case of poor bit availability, and with a step obtained by reducing said real step by a second predetermined quantity (p3) in case of high bit availability, the coding quantization step to be used in the latter case being a unitary step if the value of the real step is less than the second predetermined quantity; e) updating the real quantization step, whenever the coding is performed with a quantization step lower than the real quantization step.

Claim 2

A method of hybrid digital coding of video signals, concerning both moving and still pictures, and organized into a sequence of frames which occur at a predetermined frequency and are formed by a sequence of lines each comprising a predetermined number of pixels, wherein: each frame is subdivided into pixel groups defining a plurality of hierarchical levels; for each frame there are coded, after quantization by a quantization law chosen for each group belonging to a first hierarchical level, coefficients of a two-dimensional transform applied either to a current frame or to the meaningful differences between the current frame and a predicted frame which has been possibly submitted to a motion compensation; and a coded signal is generated which comprises, among other information, information relevant to the coding type and to the quantization law employed, and which is asynchronously transmitted at variable bit rate, the quantization law adopted for a pixel group of the first hierarchical level using a quantization step determined by a coding control device (CCD) and varying depending on transmission bit availability, wherein, for each frame, the pixel groups are counted which belong to a second hierarchical level and which have been submitted to a predetermined coding type for a first number of previous consecutive frames, to recognize whether the frame belongs to a still or to a moving picture, and upon recognition of a still picture, for each group of pixels of such a picture belonging to said first level the quantization step is made to converge towards a predetermined minimum value (p1), which is the same for all groups, by the following operations: a) memorizing the quantization steps used for that group in the different frames of the still picture; b) detecting and memorizing a real quantization step (RQS) which is the minimum quantization step used till the current frame; c) comparing the real quantization step (R) and the quantization step determined for that group in the current frame by the coding control device (CCD); d) coding the transform coefficients relevant to that group with a quantization step (CQS) equal to that determined by the coding control device (CCD), if the latter step is lower than the real quantization step (RQS) or, if the quantization step determined by the coding control device is greater than or equal to the real quantization step (RQS), coding of the coefficients by a particular quantization step obtained by increasing by a first predetermined quantity (p2) the real quantization step, in case of poor bit availability, and with a step obtained by reducing said real step by a second predetermined quantity (p3) in case of high bit availability, the coding quantization step to be used in the latter case being a unitary step if the value of the real step is less than the second predetermined quantity; e) updating the real quantization step, whenever the coding is performed with a quantization step lower than the real quantization step.