1 # libaec - Adaptive Entropy Coding library
3 Libaec provides fast lossless compression of 1 up to 32 bit wide
4 signed or unsigned integers (samples). The library achieves best
5 results for low entropy data as often encountered in space imaging
6 instrument data or numerical model output from weather or climate
7 simulations. While floating point representations are not directly
8 supported, they can also be efficiently coded by grouping exponents
12 [Golomb-Rice](http://en.wikipedia.org/wiki/Golomb_coding) coding as
13 defined in the Space Data System Standard documents [121.0-B-2][1] and
18 Source code and binary installer can be [downloaded here](https://gitlab.dkrz.de/k202009/libaec/tags).
22 In [patent.txt](doc/patent.txt) a statement on potentially
23 applying intellectual property rights is given.
27 See [INSTALL](INSTALL) for details.
31 [Libaec can replace SZIP](README.SZIP).
35 In this context efficiency refers to the size of the encoded
36 data. Performance refers to the time it takes to encode data.
38 Suppose you have an array of 32 bit signed integers you want to
39 compress. The pointer pointing to the data shall be called `*source`,
40 output goes into `*dest`.
46 struct aec_stream strm;
50 /* input data is 32 bits wide */
51 strm.bits_per_sample = 32;
53 /* define a block size of 16 */
56 /* the reference sample interval is set to 128 blocks */
59 /* input data is signed and needs to be preprocessed */
60 strm.flags = AEC_DATA_SIGNED | AEC_DATA_PREPROCESS;
62 /* pointer to input */
63 strm.next_in = (unsigned char *)source;
65 /* length of input in bytes */
66 strm.avail_in = source_length * sizeof(int32_t);
68 /* pointer to output buffer */
71 /* length of output buffer in bytes */
72 strm.avail_out = dest_length;
74 /* initialize encoding */
75 if (aec_encode_init(&strm) != AEC_OK)
78 /* Perform encoding in one call and flush output. */
79 /* In this example you must be sure that the output */
80 /* buffer is large enough for all compressed output */
81 if (aec_encode(&strm, AEC_FLUSH) != AEC_OK)
84 /* free all resources used by encoder */
85 aec_encode_end(&strm);
89 `block_size` can vary from 8 to 64 samples. Smaller blocks allow the
90 compression to adapt more rapidly to changing source
91 statistics. Larger blocks create less overhead but can be less
92 efficient if source statistics change across the block.
94 `rsi` sets the reference sample interval. A large RSI will improve
95 performance and efficiency. It will also increase memory requirements
96 since internal buffering is based on RSI size. A smaller RSI may be
97 desirable in situations where each RSI will be packetized and possible
98 error propagation has to be minimized.
102 * `AEC_DATA_SIGNED`: input data are signed integers. Specifying this
103 correctly increases compression efficiency. Default is unsigned.
105 * `AEC_DATA_PREPROCESS`: preprocessing input will improve compression
106 efficiency if data samples are correlated. It will only cost
107 performance for no gain in efficiency if the data is already
110 * `AEC_DATA_MSB`: input data is stored most significant byte first
111 i.e. big endian. You have to specify `AEC_DATA_MSB` even if your host
112 architecture is big endian. Default is little endian on all
115 * `AEC_DATA_3BYTE`: the 17 to 24 bit input data is stored in three
116 bytes. This flag has no effect for other sample sizes.
118 * `AEC_RESTRICTED`: use a restricted set of code options. This option is
119 only valid for `bits_per_sample` <= 4.
121 * `AEC_PAD_RSI`: assume that the encoded RSI is padded to the next byte
122 boundary while decoding. The preprocessor macro `ENABLE_RSI_PADDING`
123 needs to be defined while compiling for the encoder to honour this
128 The following rules apply for deducing storage size from sample size
131 **sample size** | **storage size**
134 9 - 16 bits | 2 bytes
135 17 - 24 bits | 3 bytes (only if `AEC_DATA_3BYTE` is set)
136 25 - 32 bits | 4 bytes (if `AEC_DATA_3BYTE` is set)
137 17 - 32 bits | 4 bytes (if `AEC_DATA_3BYTE` is not set)
139 If a sample requires less bits than the storage size provides, then
140 you have to make sure that unused bits are not set. Libaec does not
141 check this for performance reasons and will produce undefined output
142 if unused bits are set. All input data must be a multiple of the
143 storage size in bytes. Remaining bytes which do not form a complete
144 sample will be ignored.
146 Libaec accesses `next_in` and `next_out` buffers only bytewise. There
147 are no alignment requirements for these buffers.
151 `aec_encode` can be used in a streaming fashion by chunking input and
152 output and specifying `AEC_NO_FLUSH`. The function will return if either
153 the input runs empty or the output buffer is full. The calling
154 function can check `avail_in` and `avail_out` to see what occurred. The
155 last call to `aec_encode()` must set `AEC_FLUSH` to drain all
156 output. [aec.c](src/aec.c) is an example of streaming usage of encoding and
161 Encoded data will be written to the buffer submitted with
162 `next_out`. The length of the compressed data is `total_out`.
164 See libaec.h for a detailed description of all relevant structure
165 members and constants.
170 Using decoding is very similar to encoding, only the meaning of input
171 and output is reversed.
177 struct aec_stream strm;
178 /* this is now the compressed data */
179 unsigned char *source;
180 /* here goes the uncompressed result */
183 strm.bits_per_sample = 32;
184 strm.block_size = 16;
186 strm.flags = AEC_DATA_SIGNED | AEC_DATA_PREPROCESS;
187 strm.next_in = source;
188 strm.avail_in = source_length;
189 strm.next_out = (unsigned char *)dest;
190 strm.avail_out = dest_lenth * sizeof(int32_t);
191 if (aec_decode_init(&strm) != AEC_OK)
193 if (aec_decode(&strm, AEC_FLUSH) != AEC_OK)
195 aec_decode_end(&strm);
199 It is strongly recommended that the size of the output buffer
200 (`next_out`) is a multiple of the storage size in bytes. If the buffer
201 is not a multiple of the storage size and the buffer gets filled to
202 the last sample, the error code `AEC_MEM_ERROR` is returned.
204 It is essential for decoding that parameters like `bits_per_sample`,
205 `block_size`, `rsi`, and `flags` are exactly the same as they were for
206 encoding. Libaec does not store these parameters in the coded stream
207 so it is up to the calling program to keep the correct parameters
208 between encoding and decoding.
210 The actual values of coding parameters are in fact only relevant for
211 efficiency and performance. Data integrity only depends on consistency
217 [Consultative Committee for Space Data Systems. Lossless Data
218 Compression. Recommendation for Space Data System Standards, CCSDS
219 121.0-B-2. Blue Book. Issue 2. Washington, D.C.: CCSDS, May 2012.][1]
220 [1]: http://public.ccsds.org/publications/archive/121x0b2.pdf
222 [Consultative Committee for Space Data Systems. Lossless Data
223 Compression. Recommendation for Space Data System Standards, CCSDS
224 120.0-G-3. Green Book. Issue 3. Washington, D.C.: CCSDS, April 2013.][2]
225 [2]: http://public.ccsds.org/publications/archive/120x0g3.pdf