CN106941610B - Binary ROI mask coding method based on improved block coding - Google Patents

Abstract

The invention proposes a binary ROI mask coding method based on improved block coding, which mainly solves the problem of low coding efficiency of the existing ROI mask. The implementation scheme is: 1) divide the ROI mask into blocks, and use a symbol to represent it, and scan all the blocks to obtain a one-dimensional symbol sequence; 2) calculate the run-length array of the symbol '0' in the symbol sequence, and convert the original symbol The consecutive symbols '0' in the sequence are replaced by a symbol '0' to obtain the corrected symbol sequence; 3) Huffman encoding is performed on the run-length array and the corrected symbol sequence respectively, and the encoded code streams of the two are combined , to get the final encoded code stream. The invention has the characteristics of simple encoding process, high efficiency of symbol sequence encoding and saving a lot of code streams, and can be used in the compression process of remote sensing image ROI mask.

Description

Binary ROI mask coding method based on improved block coding

technical field

The invention belongs to the technical field of image processing, and in particular relates to a binary ROI mask coding method, which can be used to realize high-efficiency coding of a remote sensing cloud image ROI mask.

Background technique

Block coding was first used in the coding of binary images, such as fax images, and was extended to the coding of grayscale images because of the simplicity and effectiveness of the method. Block coding is divided into two processes: block scanning and Huffman coding: first select a block with a set size to divide the image, and then use the arrangement information of the gray value in the block as the message representing the block, and use The corresponding symbols are represented, and then the entire image is scanned, and the symbols corresponding to all blocks are formed into a one-dimensional symbol sequence, and finally the symbol sequence is Huffman coded.

This traditional block coding effectively utilizes the spatial correlation of binary images, but in the coding of symbol sequences, since the characteristics of symbol sequences corresponding to different binary images are not considered, the traditional Huffman coding method is always used , thus reducing the efficiency of symbol sequence encoding.

Contents of the invention

The purpose of the present invention is to address the shortcomings of the above-mentioned prior art, and propose a binary ROI mask coding method based on improved block coding, so as to improve the coding efficiency of the ROI mask.

The technical idea of the present invention is: for the region of the ROI mask and the characteristics of a large number of statistical redundancy in the symbol sequences corresponding to these regions, the ROI mask is divided into blocks by using the spatial correlation between the pixels of the ROI mask, A one-dimensional symbol sequence is obtained by symbol packaging each block, and the coding of the ROI mask is the coding of the symbol sequence. Processing, calculating its run-length array, and performing encoding of the run-length array and the corrected symbol sequence. Its specific implementation plans include the following:

(1) The ROI mask image is divided into 3*3 blocks in order from top to bottom and from left to right, and the pixel value in the block is read as a nine-bit binary 0, 1 sequence, and the binary The sequence is converted into a decimal value as a symbol representing the block; then, in the order from top to bottom and left to right, all 3*3 blocks are scanned, and all symbols are read as a one-dimensional symbol sequence;

(2) Traverse the one-dimensional symbol sequence, calculate the run-length array of the '0' symbol in the symbol sequence, and replace the continuous symbol '0' in the symbol sequence with a corresponding symbol to obtain the corrected symbol sequence;

(3) Carry out Huffman coding on the run-length array and the corrected symbol sequence respectively, and then combine the code streams of the two to obtain the final code stream.

Compared with the prior art, the present invention has the following advantages:

First, the present invention utilizes the spatial correlation of the ROI mask, reducing the statistical redundancy of the ROI mask

Second, the present invention can use a shorter code stream to complete the encoding of the ROI mask, thereby improving the encoding efficiency of the ROI mask.

Description of drawings

Fig. 1 is the realization flowchart of the present invention;

Fig. 2 is a schematic diagram of block scanning in the present invention.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings and embodiments.

Step 1, block the ROI mask to obtain a symbol sequence.

(1a) Create a two-dimensional array A, read the ROI mask of the region of interest extracted in the region of interest into the two-dimensional array A, and perform a 3*3 block operation, and follow the method shown in Figure 2 Read the pixel value in the block to get the binary sequence;

(1b) Create a variable T to store the symbol corresponding to the block, and initialize it to 0, then traverse the binary sequence from high to low, and add the decimal value corresponding to each bit in the binary sequence to T, after the traversal Get the symbol corresponding to the block;

(1c) Traversing the symbols corresponding to all blocks to obtain a sequence of symbols.

Step 2, traverse the symbol sequence to obtain the run-length array of the symbol '0' and the corrected symbol sequence.

(2a) Create a one-dimensional array B for storing the run length of symbol '0', and initialize it as empty;

(2b) Start traversing from the first symbol of the symbol sequence to determine whether the symbol is '0':

If the symbol is not '0', skip the symbol and execute (2c);

If the symbol is '0', calculate the number n of consecutive '0' after the symbol, add n to the array B, skip n symbols '0', and execute (2c);

(2c) Determine whether the end of the symbol sequence is reached:

If the end is not reached, return to (2b) to continue traversing;

If it reaches the end, the traversal ends and a run length array with the symbol '0' is obtained;

(2d) Traverse again from the first symbol of the symbol sequence to determine whether the symbol is '0':

If the symbol is not '0', skip the symbol and execute (2e);

If the symbol is '0', delete all consecutive symbols that are '0' after the symbol, skip the symbol, and execute (2e);

(2e) Determine whether the end of the symbol sequence is reached:

If the end is not reached, step (2d) is performed;

If the end is reached, the traversal ends and the corrected sequence of symbols is obtained.

Step 3, perform Huffman coding on the result obtained in step 2, and merge code streams.

(3a) Calculate the probability of each value in the run length array of the symbol '0', and allocate the code length according to the probability corresponding to each value: the greater the probability of the value, the shorter the allocated code length, and the smaller the probability, the The longer the allocated code length, finally encode all the values of the run-length array according to the allocated code word, and obtain the encoded code stream of the run-length array with the symbol '0';

(3b) Calculate the probability of each symbol in the corrected symbol sequence, and allocate the code length according to the probability corresponding to the symbol: the greater the probability of the symbol, the shorter the allocated code length, and the smaller the probability, the longer the allocated code length. Finally, all the symbols in the corrected symbol sequence are encoded according to the allocated codeword, and the encoded code stream of the corrected symbol sequence is obtained;

(3c) Add the coded code stream length of the run-length array of the symbol '0' obtained in (3a) to before the coded code stream of the run-length array, and splice the coded code stream of the corrected symbol sequence obtained in step (3b) After the coded code streams of the run-length array are reached, the code streams are merged to obtain the final coded code streams.

Taking the symbol sequence {2,0,0,0,0,511,511,5,0,0,0} as an example, the specific implementation of this step is described as follows:

First, execute steps (2a) to (2c), get the run length array of symbol '0' as [43], and execute steps (2d) to (2e) to get the corrected symbol sequence as {2,0,511,511,5,0 };

Next, step (3a) is performed to encode the run-length array. Since the two values 4 and 3 in the run-length array have the same probability, the two values are assigned the same code length, which is 0 and 1 respectively. According to the assigned The code length encodes the run-length array to obtain the encoded code stream '01'.

Next, step (3b) is performed to calculate the probabilities of symbols '2', '0', '511' and '5' in the corrected symbol sequence, which are respectively and And assign a code length of 1 to the symbols '0' and '511' with the highest probability, which are 0 and 1 respectively; assign a code length of 2 to the symbols '2' and '511' with a lower probability, respectively are 01 and 10; and then encode the corrected symbol sequence according to the allocated code length to obtain the encoded code stream '01011100';

Finally, execute step (3c), calculate the length of the encoded code stream of the run-length array as 2, and represent it with a byte 00000010, and then add this byte before the encoded code stream '01' of the run-length array to obtain the code stream '0100000010 '; After adding the code stream '0100000010' to the code stream '01011100' of the corrected symbol sequence, the code stream '010000001001011100' is finally obtained, and the encoding of the ROI mask is completed.

The above description is only a specific example of the present invention, and does not constitute any limitation to the present invention. Obviously, for those skilled in the art, after understanding the contents and principles of the present invention, it is possible without departing from the principles and structures of the present invention. Various modifications and changes in form and details are made, but these modifications and changes based on the idea of the present invention are still within the protection scope of the claims of the present invention.

Claims (4)

1. A binary ROI mask coding method based on improved block coding comprises the following steps:

(1) 3-3 partitioning the ROI mask image from top to bottom and from left to right, reading pixel values in the partitions into a nine-bit binary 0 and 1 sequence, and converting the binary sequence into decimal values serving as symbols for representing the partitions; scanning all the 3 x 3 blocks according to the sequence from top to bottom and from left to right, and reading all the symbols into a one-dimensional symbol sequence;

(2) traversing the one-dimensional symbol sequence, calculating a run-length array of '0' symbols in the symbol sequence, and replacing continuous '0' symbols in the symbol sequence with a corresponding symbol to obtain a corrected symbol sequence;

(3) and respectively carrying out Huffman coding on the run-length array and the corrected symbol sequence, and merging the code streams of the run-length array and the corrected symbol sequence to obtain a final code stream.

2. The method of claim 1, wherein the run-length array of '0' symbols in the sequence of symbols is calculated in step (2) by:

(2a) creating an array A for storing the run and initializing the array A to be empty;

(2b) starting from the first symbol of the symbol sequence, determining whether the symbol is '0':

if the symbol is not '0', skipping the symbol, performing (2 c);

if the symbol is '0', calculating the number n of consecutive '0's after the symbol, adding n to the array A, and skipping n symbols '0', performing (2 c);

(2c) judging whether the end of the symbol sequence is reached, if the end is not reached, returning to the step (2b) to continue traversing; if the end is reached, the traversal ends.

3. The method of claim 1, wherein the step (2) replaces successive '0' symbols in the symbol sequence with a corresponding one of the symbols by:

(2d) starting from the first symbol of the symbol sequence, determining whether the symbol is '0':

if the symbol is not '0', skipping the symbol, performing (2 e);

if the symbol is '0', deleting all the symbols which are '0' consecutively after the symbol, and skipping the symbol, performing (2 e);

(2e) judging whether the end of the symbol sequence is reached, if the end is not reached, returning to the step (2d) to continue traversing; if the end is reached, the traversal ends.

4. The method of claim 1, wherein step (3) is performed by:

(3a) calculating the probability of each value in the run-length array of the symbol '0', and distributing the code length according to the probability corresponding to the value: the larger the probability of the value is, the shorter the distributed code length is, the smaller the probability is, the longer the distributed code length is, and finally, all values of the run-length array are coded according to the distributed code words to obtain a coded code stream of the run-length array of the symbol '0';

(3b) calculating the probability of each symbol in the corrected symbol sequence, and distributing the code length according to the probability corresponding to the symbol: the larger the probability of the symbol is, the shorter the allocated code length is, the smaller the probability is, the longer the allocated code length is, and finally, all symbols in the corrected symbol sequence are coded according to the allocated code words to obtain a coded code stream of the corrected symbol sequence;

(3c) and (3) adding the length of the code stream of the run-length array of the symbol '0' obtained in the step (3a) to the front of the code stream of the run-length array, splicing the code stream of the corrected symbol sequence obtained in the step (3b) to the code stream of the run-length array, and then completing the merging of the code streams to obtain the final code stream.