ADVANCE INFORMATION
ZR36050
CODE VOLUME CONTROL
The JPEG Baseline encoding algorithm does not inherently
provide a means of achieving a target code volume for a com-
pressed image (or an equivalent target average coding rate in
bits per pixel, or target compression ratio). In fact, since the
standard specifies that the whole image frame is encoded using
the same set of quantization tables and Huffman coding tables,
it provides no built-in mechanism to modify the coding rate once
encoding of an image has started.
correct value, the Bit Rate Control operates without affecting the
compressed image quality. The ZR36050 has a Statistical Pass
mode of operation, in which it performs a preliminary pass over
the image data, and determines appropriate values of the Scale
Factor and Allocation Factor for use in a Compression Pass with
Bit Rate Control. The host specifies the target code volume to be
used in the computation of the Scale Factor and Allocation
Factor, in the TCV_DATA register. There is also a mode of oper-
ation in which the ZR36050 performs a Statistical Pass followed
automatically by a Compression Pass with Bit Rate Control. For
more details on these modes of operation, see the “ENCODING
MODES” section.
One way to achieve a target code volume in compression of a
particular image is to iterate the compression, using different
quantization tables each time, until the desired target is
achieved. The Scale Factor (SF register in the Internal Memory)
allows the host to modify the quantization tables by changing a
single parameter. In operation, the ZR36050 multiples the quan-
tization tables, specified in the DQT segment of Internal Memory,
by the Scale Factor before storing them in the Quantization
Tables Store.
The code volume overflow detection option (activated by the
OVF bit of the OPTIONS register) is useful if Bit Rate Control is
not used, or if the Scale Factor and Allocation Factor contain
inappropriate values. With this option selected, the ZR36050
continuously compares the accumulated code volume with a
target limit specified in the TCV_NET register. If the code volume
exceeds the limit at any time during a compression pass, the
TCVOVF status register bit is set, and the interrupt pin is activat-
ed to notify the host. Since this is considered a destructive event,
the ZR36050 aborts the compression pass and goes Idle.
When it is activated with appropriate values of the Scale Factor
and an Allocation Factor (AF register), the unique Bit Rate
Control feature of the ZR36050 ensures that the actual code
volume achieved is equal to or slightly less than the target code
volume. Moreover, if the Scale Factor is close enough to the
COMPRESSED DATA FORMAT
The ZR36050 supports encoding and decoding of all three com-
pressed data format classes defined by the JPEG standard:
Compressed data generated by the ZR36050 fully complies with
the JPEG standard and includes all headers and marker
segments necessary to allow it to be decoded by a compliant
JPEG decoder. The JPEG standard allows virtually unlimited
permutations in the order and repetition of optional marker seg-
ments. As a practical matter, there are some restrictions on the
permutations and repetitions in compressed data generated by
the ZR36050:
■ The full interchange format, in which the specifications of all
the tables used for encoding the image are included with the
compressed image data, in their marker segments. This
format is used when the compressed image must be
decoded by a JPEG decoder that has no knowledge of the
tables used.
■ The abbreviated format for compressed data. This format
contains the compressed image data and the frame and scan
headers, but where some or all of the tables are omitted.
■ The order of optional marker segments is fixed, as follows:
APP, COM, DRI, DQT, DHT.
■ Only one instance of each optional marker segment can be
■ The abbreviated tables-only format. This format contains
marker segments with table specifications but no
compressed image data.
inserted before each SOS, or the SOF, marker.
■ The frame header can specify up to eight image
components.
The two abbreviated formats complement each other in applica-
tions based on the ZR36050. For example, when a sequence of
image frames, such as a video clip, are compressed, the same
Huffman tables are used for all the images in the sequence. If the
compressed sequence is to be sent to a JPEG decoder that has
no prior knowledge of the tables, it is most economical for the
encoder to create initially, an abbreviated format compressed
data-less image, containing only the Huffman tables, followed by
a sequence of compressed images in which the Huffman tables
are omitted. To expand the image sequence, the JPEG decoder
first decodes the tables-only data, thus installing the Huffman
tables, which are subsequently used in the decoding of all the
images in the sequence.
■ A compressed image can contain up to eight scans.
A compressed image to be decoded by the ZR36050 must be in
JPEG standard format. The ZR36050 recognizes and decodes
all the following marker types: SOI, SOF (SOF0 or SOF3), SOS,
APP, COM, DRI, RST, DQT, DHT, DNL, EOI. All marker
segments other than JPEG Baseline marker segments, i.e.,
markers followed by a length parameter, are disregarded and do
not cause any error. Markers without a following length parame-
ter, except for SOI, RST and EOI, will cause unpredictable
behavior. A marker can be prefixed by any number of FF bytes.
In its decoding modes, the ZR36050 can expand any JPEG
Baseline compressed image. The finite size of the Internal
10
5秒后页面跳转
深入解析CS8416引脚说明、参数、技术特点
TPA3116D2功放芯片参数详解、引脚说明
移位寄存器74HC165引脚说明、驱动程序示例解读
深入解析AD9833：DDS频率合成器的卓越性能与广泛应用
微信公众号
抖音公众号
浙公网安备 33010502006866号 浙ICP备10014259号-119
营业执照ICP证
技术参数
数据手册
如果您发现信息不准确，
