NUC970 Technical Reference Manual
Publication Release Date: Dec. 15, 2015
- 1044 -
Revision V1.30
NUC97
0
T
E
CHNIC
A
L
RE
F
E
RE
N
CE
MA
NUA
L
Encode Begin
Encode End
Thumbnail
Encode
Primary
Encode
Primary
Encode
Thumbnail on
Thumbnail off
Idel State
Encode On
Figure 5.29-4 Primary and thumbnail encode
The JPEG Codec supports the planar format up-scaling and down-scaling function to adjust the
encoded image size. The primary image encode supports 1~8X arbitrarily up-scaling in horizontal and
vertical
direction. But the thumbnail image encode doesn’t support the up-scaling function. The JPEG
Codec also supports the planar format down-
scaling function with the 1/2, 1/3, 1/4, …, 1/64 ratio in
vertical direction, and 1/2, 1/4, 1/6, 1/8, …, 1/62, 1/64 ratio in horizontal direction. The programmer
can set the registers JPSCALU, JPSCALD, JTSCALD for image scaling up or down. The registers
JPRIWH and JTHBWH specify the width and height of the encoded image after scaling. For planar
format up-scaling mode, the programmer needs to specify the up-scale ratio by registers JUPRAT and
the source image height register JSRCH. The rotation function is only supported for planar format. The
source image can be encoded by rotate left or right 90°. It should be mentioned that the rotation
function can only be applied only when encode YCbCr 4:2:0 source.
The standard JPEG bit-stream format includes some headers that specify some information, For
example, Quantization-table (QTAB) and Huffman-table (HTAB) belong to the part of the header. The
JPEG Codec supports some options whether you can insert these headers into the JPEG bit-stream
or not. These options are defined in the control register JHEADER. The JPEG codec also supports
quantization-table adjustment to control the size of JPEG bit-stream. When the bit-stream size is too
large, it can set or adjust the value of quantization-table to reduce the bit-stream size. The adjustment
control is defined in registers JPRIQC and JTHBQC. In addition, three programmable quantization-
tables are provided. The programmer can specify using two or three tables for encoding by register bit
E3QTAB (JITCR[3]). The Quantization-table registers are defined in JQTAB0~JQTAB2.
The JPEG Codec supports two coding modes: single mode and continue mode. For single mode, the
programmer can use dual-buffer or fix-buffer to store JPEG bit-stream in frame memory. For continue
mode, the programmer can store the neighbor JPEG bit-streams into frame memory continuously or
store each JPEG bit-stream into frame memory by creating same buffer size that is specified in
register JFSTRIDE. The following figure specifies these two encode modes.
