CN111862247A - Enhanced dominant quadtree spatial data structure and construction method thereof - Google Patents
Enhanced dominant quadtree spatial data structure and construction method thereof Download PDFInfo
- Publication number
- CN111862247A CN111862247A CN201910357775.7A CN201910357775A CN111862247A CN 111862247 A CN111862247 A CN 111862247A CN 201910357775 A CN201910357775 A CN 201910357775A CN 111862247 A CN111862247 A CN 111862247A
- Authority
- CN
- China
- Prior art keywords
- block
- sub
- node
- attribute
- dominant
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010276 construction Methods 0.000 title claims abstract description 11
- 238000007906 compression Methods 0.000 claims abstract description 24
- 230000006835 compression Effects 0.000 claims abstract description 24
- 238000000034 method Methods 0.000 claims abstract description 12
- 230000003121 nonmonotonic effect Effects 0.000 claims description 35
- 230000007246 mechanism Effects 0.000 claims description 9
- 238000004458 analytical method Methods 0.000 claims description 7
- 238000013500 data storage Methods 0.000 claims description 4
- 230000008901 benefit Effects 0.000 abstract description 8
- 238000013144 data compression Methods 0.000 abstract description 3
- 238000012732 spatial analysis Methods 0.000 abstract description 2
- 238000011496 digital image analysis Methods 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 10
- 238000012545 processing Methods 0.000 description 7
- 230000006872 improvement Effects 0.000 description 6
- 229910002056 binary alloy Inorganic materials 0.000 description 5
- 230000007547 defect Effects 0.000 description 2
- 230000008520 organization Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
- G06T9/40—Tree coding, e.g. quadtree, octree
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an enhanced advantageous quadtree structure and a construction method thereof, and aims at efficient lossless compression and storage of high-resolution remote sensing digital images and high-precision grid graphic data. The invention also considers 28And 212The method has the advantages that gray level bright-level grid type spatial data are efficiently and losslessly compressed and stored, the storage potential of a dominant quad-tree spatial data structure is fully mined, a tree node pointer domain is transformed into a pointer-attribute complex which gives consideration to the requirements of storage addresses and storage attributes, when the corresponding image blocks of tree nodes are only composed of different monotonic sub-blocks, the attribute values of the sub-blocks are stored, the tree depth is reduced, and meanwhile, the storage efficiency is improved. Compared with the dominant quad-tree, the enhanced dominant quad-tree has the advantages that the data compression efficiency is improved remarkably; compared with a linear quadtree, the method has more advantages in compression efficiency and access efficiency, is very suitable for storing high-resolution complex remote sensing image data and large-scale grid type spatial data, and can provide powerful support for digital image analysis and grid type spatial analysis.
Description
Technical Field
The invention belongs to the technical fields of geographic information systems and remote sensing, geographic space data organization, space data structures, digital image compression storage, graphics, space big data and the like, can be applied to organization and efficient lossless compression storage of massive grid-type space data and digital images, and provides support for analysis and processing of high-resolution remote sensing image data and large-scale high-precision grid-type space data.
Background
In a storage structure of digital images and GIS grid-type spatial data, a quadtree structure is a spatial data structure with great application potential, and the structure simultaneously considers the requirements of compressed storage and efficient access. For a conventional quadtree structure, nodes of the quadtree structure are composed of an attribute field and four child node pointer fields, the conventional quadtree structure has advantages in spatial data retrieval, and can achieve both good spatial complexity and time complexity when storing general simple image data. When storing complex image data, especially large-scale complex image data, the negative compression storage performance of the image data makes it impractical.
Problems with conventional quadtrees: 1) compared with the attribute storage space for the monotonic sub-blocks in the image, the storage overhead of the node pointer field in the tree is very huge, which is usually 8 to 16 times of that of the attribute storage space for the monotonic sub-blocks in the image; 2) 1/4 nodes in the tree are used as intermediate path nodes, and the attribute fields of the nodes are wasted in an idle mode; 3) the pointer value of the leaf node occupying the tree summary point 3/4 is null, which is a main reason for low compression storage efficiency; 4) when storing complex images, it often appears as a negative compression, which is not suitable for storing complex images, high resolution images and large scale raster data.
The linear quad-tree structure is an improvement of the conventional quad-tree structure, the structure takes attribute information stored by all leaf nodes in the conventional quad-tree together with path information for identifying the spatial positions of the nodes as a storage target, and stores the attribute information and the path information in the form of a linear table, pointer spaces of all nodes of the conventional quad-tree are released, and the improvement is mainly embodied in the relative improvement of data compression efficiency, but at the cost of abandoning data expression of the spatial quad-tree structure.
Related researches are focused on the aspects of improvement of a construction algorithm of a linear quadtree and the like, and the aims of improving construction efficiency, overcoming low data access efficiency and enhancing operability and practicability of the linear quadtree are fulfilled. For the construction of linear quadtrees, a construction method based on recursive decomposition, a linear quadtree coding mode of two-dimensional run-length coding and an optimized quadtree construction algorithm by means of a digital search tree appear, and in order to further improve the efficiency of constructing the linear quadtree, researchers provide linear quadtrees based on decimal Morton codes and various improved code constructing methods thereof, such as linear quadtree structures of variable-length path coding and the like.
There are also some problems with the linear quadtree structure: 1) the efficient space search algorithm based on the quad-tree structure cannot be directly utilized due to discarding of intermediate path nodes of the conventional quad-tree; 2) because the node attribute field stores only the attribute of a monotone sub-block, the number of nodes to be stored is still huge, which accounts for 3/4 of the total number of the nodes of the conventional quadtree, the nodes also need to store path codes for identifying the spatial positions of the nodes, and the compression efficiency is still limited; 3) in view of the fact that the spatial analysis processing oriented to high-precision complex digital images and high-precision large-scale grid patterns heavily depends on high-frequency random retrieval and access of data, the application of a linear quad-tree structure is restricted.
Aiming at overcoming the problems and the defects of the conventional quadtree structure and the linear quadtree structure, 28In 2008, authors proposed a quadtree space data structure based on dominant attribute storage, referred to as a dominant quadtree structure for short, for lossless compression storage of gray-scale and bright-scale images. In the dominant quadtree, the node structure consists of a feature code, an attribute field, and a child node table pointer. The characteristic code is unsigned character type data, the child node table pointer occupies 4 bytes, and the dominant attribute value field is an unsigned short integer of a single byte.
The advantageous quadtree structure is:
the single byte feature code data in the advantageous quadtree structure is an 8-bit binary number (8bit, see fig. 1), and is composed of three information segments, which are respectively used to describe the Position (Position) of the current node corresponding graph block in the parent block (the last layer of graph block), the number of children (sub-block) Nodes (Nodes) that the current node needs to be extended, and the advantageous attribute monotonic sub-block distribution Template (Template) in the current node corresponding graph block, and the feature codes and each information segment are defined as follows:
first 2-bit binary number d in dominant quadtree node feature code0d1Encoding the position of the current image block in the parent block; intermediate 2-bit binary number d 2d3The number of Nodes corresponding to child Nodes of the current block subblock is; last 4 binary digits d4d5d6d7The Template is distributed for the monotonic sub-block dominance attribute of the current tile.
Fig. 2 shows position numbers of 4 sub-blocks obtained by quartering a parent block, where the upper left is a first sub-block, the upper right is a second sub-block, the lower left is a third sub-block, the lower right is a fourth sub-block, and binary position codes 00, 01, 10, 11 or decimal position codes 0, 1, 2, 3 respectively indicate that the current block is located at the upper left, upper right, lower left, and lower right positions of the parent block.
In the superior quadtree node characteristic information code, when the binary value information segment node which represents the number of child Nodes of the current node is d2d301, 10, 11, respectively indicate that there are 1, 2, 3 child nodes in the current block.
In the superior quadtree node characteristic information code, when the binary value information segment node which represents the number of child Nodes of the current node is d2d3The meaning of 00 can be interpreted according to 2 cases divided by the binary value of Template: when the Template is 1111, the current node is completely monotonous corresponding to the image block, no sub-block node exists, namely the number of child nodes is 0; when the Template is 0000, it indicates that there are 4 non-monotonic sub-block images corresponding to the current node corresponding to the tile, meaning that there are 4 child nodes.
The advantage attribute refers to the common attribute of the monotonous subblocks with the same attribute value, which are dominant in quantity, in the case that the monotonous subblocks exist in the current block, and the advantage attribute value is stored in the attribute/gray level domain of the current node.
The dominant attribute sub-block distribution Template is used for recording and describing distribution characteristics of a plurality of monotonic sub-blocks with dominant attributes in a current image block, the binary range of values of the distribution Template is 0000-1111, and the hexadecimal range of values of the distribution Template is 0-F.
If the current image block does not have the monotone sub-block, the dominance attribute sub-block hexadecimal distribution characteristic Template of the corresponding node is 0.
If the current block has monotonic sub-blocks, the dominant attribute sub-block distribution characteristic Template value of the corresponding node can be illustrated with reference to fig. 3, where the 16- ary Template values 1, 2, 4, and 8 are basic Template values, which respectively indicate that one dominant attribute monotonic sub-block is located at 4 positions of the block, i.e., upper left, upper right, lower left, and lower right, and the others are combined templates, which indicate the distribution characteristics of a plurality of same-valued dominant attribute monotonic sub-blocks in the current block.
Phase contrast is to 28For a gray level image storage conventional quad-tree structure, the advantageous quad-tree structure adds 1 single-byte characteristic information code, and an attribute domain is expanded from an attribute value for storing a current monotone block to a common attribute value capable of storing a plurality of monotone sub-blocks with the same value, so that the depth of the quad-tree is effectively reduced, the total number of nodes is greatly reduced, the proportion of information nodes is improved, and the utilization efficiency of the attribute domain is improved.
The number of the children pointers of the superior quadtree is reduced from 4 in the conventional quadtree structure to 1, namely, only 1 pointer points to the first address of the child node sequence table (the length is less than or equal to 4), the byte length of the node is reduced from 17(1+4 multiplied by 4) bytes of the conventional quadtree node to 6(1+1+4) bytes, and the reduction amplitude is 64.7%.
The two improvements of the advantageous quadtree greatly improve the image or raster data compression storage performance of the quadtree compared with the conventional quadtree, and show the practicability with great prospect.
In the advantageous quadtree, if the current tile is not monotonic and its 4 sub-tile images are not monotonic, the signature of its corresponding node can be illustrated by the example in fig. 4. As shown, the current tile image is not monotonic and is in the upper right-hand position of its parent tile image, its position is encoded as binary 01 (01)2The current block is composed of 4 sub-blocks with non-monotonous attributes, 4 sub-blocks correspond to 4 child nodes, and the number of child nodes is 4, namely binary (100)2Taking the last two digits (00)2Since no dominant attribute value is stored, the dominant attribute distribution template is (0000)2. So that the feature code of the node corresponding to the current image block is determined by its position (01) in the parent block2Node number of sub-block (00) 2And dominant property distribution template (0000)2Composition, namely (01000000)2I.e., hexadecimal 40; the pointer field stores a first address that points to the child node table.
In the dominant quadtree, if the current tile is completely monotonic, i.e. it consists of 4 monotonic sub-block images with the same attribute value, the feature codes of the corresponding nodes can be explained by way of example in fig. 5. The binary system of the 16-system feature code 0F is 00001111, wherein the current image block is positioned at the upper right part in the father block, the position code is 00, the number of child nodes is 00 because of monotony, the child nodes do not need to be constructed, the child nodes and the child nodes form the first 4 bits 0000 of the feature code, the distribution templates of four monotone sub-blocks with the same value are 1111 and the 16-system F, and the node pointer field is null value because no child node can point.
In the dominant quadtree, if the current tile contains monotonic sub-blocks, the signatures of the corresponding nodes can be illustrated by the two examples in FIG. 6. Wherein sub-graph (a) shows a tile case containing both monotonic and non-monotonic sub-blocks; sub-graph (b) shows the case of a tile containing 4 monotonic sub-blocks that are not exactly identical.
As shown in FIG. 6(a), if the current block image is not monotonous and is in the lower left corner of its parent block image, its position is coded as binary 10 (10) 2The current block is composed of 3 monotone sub-blocks and one non-monotone sub-block, wherein 00 blocks and 11 blocks are two dominant attribute sub-blocks with the same value, 01 blocks are attribute non-monotone sub-blocks, and 10 blocks are non-dominant attribute sub-blocks. The attribute domain of the node corresponding to the current block stores the common dominant attribute gray value of the two sub-blocks, and the attribute domain has two child nodes which respectively correspond to the non-monotonic sub-blocks and the monotonic non-dominant attribute sub-blocks; the feature code is determined by its position in the parent block (10)2Node number of sub-blocks (10)2And advantageous Property distribution template (1001)2Composition, namely (10101001)2I.e., hexadecimal A9, the pointer field holds a pointer to the childThe first address of the sub-node table. The feature codes, attribute values and pointers of the child nodes corresponding to the non-monotonic sub-blocks are determined by the image features of the non-monotonic sub-blocks, and the determination rule is consistent with the current node; the node corresponding to the monotonic non-dominant attribute sub-block is a leaf node, and the hexadecimal feature code thereof is 8F (10001111)2Indicating that it is in the lower left corner of the parent tile (10)2Position, the number of points of the child block is 0 (00)2The dominant monotone sub-block distribution template is (1111)2The pointer field is null.
As shown in fig. 6(b), if the current tile is located at the lower right corner of the parent tile, its location is encoded as binary 11, which is composed of 4 monotonic sub-blocks, where 01 block and 10 block are two dominant attribute sub-blocks with the same value, and 00 block and 11 block are attribute non-dominant monotonic sub-blocks. The attribute domain of the node corresponding to the current image block stores the dominant attribute gray value, and has two child nodes which respectively correspond to two non-dominant monotone attribute sub-blocks; the feature code is determined by its position in the parent block (11) 2Node number of sub-blocks (10)2And advantageous Property distribution template (0110)2Composition, namely (11100110)2I.e., hexadecimal E6, the pointer field holds the first address that points to the child node table. A monotonic non-dominant attribute sub-block node feature code 0F (00001111)2Indicating that it is located at the 00 position of the current block and the node number of the sub-block is 0 (00)2And single template (1111)2The pointer field is null; another monotonic non-dominant attribute sub-block node feature code (11001111)2I.e., hexadecimal CF, indicating that it is in the parent tile (11)2Number of position, sub-block nodes (00)2And single template (1111)2The pointer field is null.
Through intensive research on the superior quadtree structure and the compressed storage mechanism thereof, the inventor finds that although the superior quadtree structure achieves great performance improvement compared with the conventional quadtree structure and the linear quadtree, the utilization form of a single pointer field with 4 bytes is single, and the single pointer field is only used as the first address for storing the child node table. When a current node is corresponding to a non-monotonic block consisting of 4 monotonic sub-blocks, extended child nodes need to be stored for the non-dominant monotonic sub-blocks, whether further extension of the nodes can be terminated by utilizing polymorphism of a pre-established pointer field storage space or not is judged, a plurality of monotonic sub-block attribute values are directly combined and stored, the depth of the 1-layer dominant quad-tree is substantially reduced, and a space for further improving and improving the storage performance still exists, so that a chance is provided for the key technology of the enhanced dominant quad-tree provided by the invention.
Disclosure of Invention
The invention is directed to 28And 212The method aims to deeply optimize and expand the dominant quadtree data structure and further enhance the lossless compression performance and the applicability of the dominant quadtree for storing digital images and raster data.
The invention also aims to provide a method for constructing an enhanced advantageous quadtree data structure, so as to overcome the defects of low compression efficiency and poor practicability of the conventional quadtree structure on digital image and raster data storage; the linear quadtree structure has low comprehensive efficiency on the storage and access of large-scale complex digital images or raster data; the advantageous quadtree does not utilize the node pointer domain to the maximum.
Another object of the present invention is to provide an access method for digital images and raster data stored in an enhanced advantageous quadtree structure, so as to solve the problem of inefficient access of images and raster data stored in a high-efficiency lossless compression data structure.
To achieve the above objects, the present invention designs and proposes an enhanced advantageous quadtree data structure different from the advantageous quadtree structure. For storage 2 8Grayscale bright level image and 212The gray level and bright level images respectively give corresponding spatial data structures so as to meet the requirement of efficient lossless compression and storage of different gray level and bright level images or raster data.
When the enhanced advantageous quadtree data structure is oriented to storage 28When the gray level image is in a bright level, the definition of the tree node data type is described as follows:
in the above data structure type definition, the pointer-attribute association is used as a pointer domain for storing the node address of the sub-block image expanded by the current node only when the non-monotonic sub-block image exists in the image block corresponding to the current node.
If the image block corresponding to the current node is composed of 4 monotone sub-block images with incompletely same values, the current node is not extended downwards any more, but is used as a leaf node, and the pointer-attribute association is used as a storage unit to store the attributes or gray values of the 4 monotone sub-block images respectively.
When the enhanced advantageous quadtree data structure is oriented to storage 212When the gray level image is in a bright level, the node data structure is described as follows:
in the data structure type definition, the pointer-attribute complex is used as a pointer domain for storing the node address of the sub-block image expanded by the current node only when the non-monotonic sub-block image exists in the image block corresponding to the current node;
If the image block corresponding to the current node is composed of 4 monotone sub-block images with incompletely same values, the current node is not extended downwards and is a leaf node, at the moment, the pointer-byte attribute array complex respectively stores the last 8-bit binary data of the gray values of the 4 monotone sub-blocks, and the first 4-bit binary data of the gray values of the 4 monotone sub-blocks are spliced into 16-bit binary data to be stored in the gray/attribute domain of the node.
The enhanced advantageous quadtree node feature code and the advantageous quadtree node feature code are designed basically in the same way, and are still unsigned single byte fields containing 3 information segments, and the binary expression is d0d1d2d3d4d5d6d7，d0d1Location, d, of the current tile in the parent tile for the current tile representation of the node2d3Is the child node of the nodeCounting Nodes; d4d5d6d7And distributing the Template for the dominant attribute of the monotone sub-block in the current block.
The difference between the above feature code and the dominant quadtree node feature code lies in the binary information segment d of the node number of children2d3When the value is 00, the meaning can be determined according to the corresponding binary Template4d5d6d7The value of (2) is 3 case interpretations, when the Template is 1111, the current block is monotonous, no sub-block node exists, and the number of child nodes is 0; when the Template is 0000, it indicates that there are 4 (binary 100) non-monotonic sub-block nodes in the current tile; when the Template is greater than 0000 and the Template is less than 1111, it indicates that the current tile includes 4 non-homonymous monotonic sub-blocks (the sub-blocks outside the domain are regarded as 0 gray-level monotonic sub-blocks), the attributes/gray-levels of these monotonic sub-blocks are stored in the attribute domain and the pointer domain of the current node, and there is no sub-block node.
The gray level/attribute domain of the strong dominant quadtree node stores dominant monotonic gray levels or attributes when the current node corresponding to the tile contains both monotonic and non-monotonic sub-blocks, the non-monotonic sub-blocks and the non-dominant monotonic sub-blocks are represented by the extended child nodes and pointed to their first addresses by the child pointers.
The gray level/attribute domain of the enhanced advantageous quadtree node is empty when the current node corresponding image block only contains 4 non-monotonic sub-blocks, and the child pointer points to the head address of the 4 child node tables extending downwards from the current node.
The method is realized by constructing a main function and constructing a recursive function by using sub-block nodes called by the main function, and adopts a top-down recursive quadrilateration mechanism and a bottom-up four-neighborhood sub-block feature analysis and node construction regression mechanism.
The constructed master function is 2n×2nThe normalized image/grid is used as a graphic input parameter, the subblock information structure variable is used as an output parameter, subblock nodes are called to construct a recursion function, the child node table initial address value returned by the recursion function is given to a child pointer of a root node, and the analysis of the characteristic information parameters of 4 subblocks is carried outAnd assembling the feature codes of the root nodes, and if dominant monotonous subblocks exist, giving attribute values of the feature codes to a gray level/attribute domain.
The method for constructing the recursive function by the enhanced dominant quadtree sub-block nodes mainly comprises the following key steps:
firstly, judging the property of a current image block, if the current image block is a bottom layer image element image block, judging the current image block to be a monotone image block, obtaining an image element attribute value and a monotone sub-block distribution hexadecimal template value F (binary value 1111) by referring to information parameters, and returning to NULL;
if the size of the current image block is larger than 2 multiplied by 2, the current image block is divided into 4 sub-block images, the function is called for the 4 sub-block images in a recursion mode respectively, and feature information of the 4 sub-block images and the head address of a possible sub-block node table of the next layer are obtained;
according to the analysis of the characteristic information parameters of the 4 subblock images, acquiring dominant monotonic subblock image attributes and distribution template information of a current image block, and the number, positions and monotonic values of non-monotonic or non-dominant subblock image nodes to be stored;
if the distribution template value of the sub-block of the current image block advantage attribute is hexadecimal F (binary value 1111) or the node number of the sub-block image is 0, the current image block is monotonous and returns to NULL;
if the current block only consists of the monotone sub-blocks with non-identical values, the corresponding nodes are leaf nodes, the low 8-bit binary system of the gray values of the 4 monotone sub-blocks is organized in the attribute-pointer combination, and the storage 2 is paired 12The gray level bright level image is characterized in that the high 4 bits of the gray values of 4 monotonic sub-blocks are assembled into a 16-bit binary number to be stored in an attribute domain, and a pointer-attribute association value is returned;
if the current image block contains non-monotonous or non-dominant sub-block images, a sub-block node table with corresponding length is created according to the number of the sub-block images, and the following assignment processing is carried out on the corresponding node of each sub-block;
if the node is a non-monotonic sub-block, the feature codes of the corresponding nodes are obtained by splicing according to the positions of the node in the father block, the number of child nodes and a monotonic dominant sub-block distribution template, and attribute values of the monotonic dominant sub-blocks and child node table head address-attribute value complex values are given to the corresponding nodes;
if the corresponding node is a non-dominant monotonic sub-block, the corresponding node is a leaf node, the number of child nodes is 0, the feature code of the corresponding node is obtained by splicing according to the position of the corresponding node in the father block and a monotonic dominant sub-block distribution template, the attribute value of the monotonic dominant sub-block is given to the corresponding node, and the child pointer is given with NULL;
and returning the first address of the node table of the sub block of the current block.
The invention provides a method for reading specified pixel or unit attributes in digital image or raster data stored in an enhanced dominant quadtree, which is realized by reading a function. Assuming that the Size of the normalized digital image or the grid data is Size, the root address of the corresponding enhanced dominant quadtree is AQT, and the row and column positions of the read pixels/units are (i, j);
The reading function adopts a quick searching mechanism based on the quadtree, firstly, a current graph block node pointer T is set to point to a root node of the enhanced advantageous quadtree, namely T is AQT, nodes stored in the appointed pixel or grid unit attribute are searched and positioned through the quadtree partitioning, and the values are completed, and the specific searching method comprises the following steps:
determining the number of the subblocks in the current image block where the designated pixel or the grid unit is located according to the 4 subblock sizes Size ═ Size/2 of the image block corresponding to the current node T and the row and column positions of the designated pixel or the grid unit;
if the current block is monotonous dominant sub-block distribution template is hexadecimal F (1111)2If the current image block is a monotone image block, directly returning the gray level/attribute value in the current node attribute domain;
if the dominant monotonic sub-block distribution template value of the current block is hexadecimal 0(0000)2If the current image block only consists of non-monotonic image blocks, the node pointer T corresponding to the current image block enters the corresponding node of the sub-block where the appointed pixel/unit (i, j) is located, and the image block pointed by the current pointer T is searched continuously;
if the dominant monotonic sub-block distribution template value of the current block is between 1 and E in hexadecimal, determining whether the designated pixel/unit falls within the dominant monotonic sub-block distribution range or not according to the dominant sub-block distribution template and the number of the sub-block where the designated pixel/unit is located, and adopting a corresponding access mechanism;
If the designated pixel/unit is in the dominant monotonic sub-block distribution range, directly returning the dominant gray/attribute value stored in the current node attribute domain;
if the current node is a leaf node, the current block is composed of only monotone sub-blocks with incompletely same value, for 28The gray level bright level image directly returns the byte value of the corresponding component in the pointer-attribute complex according to the number of the sub-block where the specified pixel/unit is located, and 2 is compared with the byte value of the corresponding component in the pointer-attribute complex12The grayscale brightness level image is assembled by the corresponding 4-bit binary value in the attribute field and the 8-bit binary value of the corresponding component in the pointer-attribute association into a 12-bit grayscale/attribute value, which is returned.
If the current node is a non-leaf node, updating the node pointer T of the current block into the node corresponding to the sub-block where the pointer T of the current block is located according to the position of the node corresponding to the non-monotonic sub-block or the non-dominant monotonic sub-block where the specified read pixel/unit is located in the node table of the sub-block of the current block, and continuously searching the block pointed by the current pointer T.
Drawings
FIG. 1 is a diagram of an advantageous quadtree signature segment;
FIG. 2 is a schematic diagram of dominant quadtree signature code sub-block position coding;
FIG. 3 is a schematic diagram of a 16-system distribution template of a monotonic dominance attribute sub-block;
FIG. 4 is a schematic diagram of a corresponding advantageous quadtree node structure with tiles that contain only non-monotonic sub-blocks;
FIG. 5 is a schematic diagram of a dominant quadtree node structure corresponding to a monotonic pattern block;
FIG. 6 is a schematic diagram of a corresponding advantageous quadtree node structure of a block containing monotonic sub-blocks;
FIG. 7 is a schematic diagram of a node structure of an enhanced dominant quadtree with only monotonic subblock tiles;
FIG. 8 is a schematic diagram of the junction structure of the enhanced dominant quadtree with only non-monotonic and dominant monotonic sub-blocks;
FIG. 9 is a schematic diagram of the junction structure of the enhanced dominant quadtree with non-monotonic and non-dominant monotonic sub-blocks;
FIG. 10 is a schematic table of enhanced advantageous quadtree compression storage performance versus other quadtrees.
Detailed Description
28Or 212An enhanced dominant quadtree of gray-level and bright-level digital image/raster data is created, and digital images are hereinafter referred to as digital image/raster data for short.
For 28Grayscale bright-level digital image using [0031]The data structure creates an enhanced dominance quadtree, for 212Grayscale bright-level digital image using [0034]The data structure creates an enhanced dominant quadtree.
Firstly, normalizing a to-be-compressed storage image/raster data with row and column size of r multiplied by c to obtain 2 n×2nNormalizing the image of (1) to 2n-1＜max(r，c)≤2n。
Establishing an enhanced superior quadtree root node, wherein the node corresponds to the whole normalized image to be compressed, calling an enhanced superior quadtree sub-block image node table by taking the normalized image as a graphic input parameter to construct a recursive function, returning the created sub-block image node table head address by the recursive function, and assigning the sub-block image node table head address to a child pointer of the root node.
According to the analysis of the characteristic information parameters of 4 sub-block images of the whole normalized image, assembling the characteristic codes reflecting the characteristics of the normalized image and assigning the characteristic codes to the characteristic code domain of the root node, and if the dominant monotonic sub-block exists, assigning the dominant attribute value to the gray level/attribute domain of the root node.
And the sub-block image node table is used for constructing a recursive function, and whether the sub-block image node table is established or not is determined by analyzing and judging the characteristics of the current image block.
If the current image block is a bottom-layer pixel image block, the current image block is used as a monotonous image block, a subblock image node table is not established, an address value NULL of the empty subblock image node table is returned, and a monotonous subblock attribute value and a monotonous subblock distribution hexadecimal template value F (binary value 1111) are obtained by referencing parameters;
if the current image block only consists of non-completely homomorphic monotonic sub-blocks, the corresponding nodes are leaf nodes, and a sub-block image node table is not established. To 2 8The gray scale digital image is characterized in that 32 bits of 8-bit binary system of 4 monotonic sub-block gray values are assembled in an attribute-pointer association, and 2 bits are combined12The gray scale digital image organizes the low 8-bit binary system of 4 monotonic sub-block gray values in the attribute-pointer association to be 32 bits in total, the high 4 bits of the 4 monotonic sub-block gray values are assembled into a 16-bit binary system to be stored in the attribute domain, and the pointer-attribute association value is returned;
for 2 stored in the above case8The gray level bright level image is explained with reference to fig. 7(a), in which the current image block is located at the upper right corner (binary 01 position) of the parent image block and is composed of 4 monotone sub-blocks with incompletely identical values, wherein 00 blocks and 11 blocks are two identical value dominant attribute sub-blocks, 01 blocks and 10 blocks are non-dominant attribute sub-blocks and no sub-block node, the current node attribute domain stores dominant attribute values, and the pointer-sub-block attribute complex stores attribute values of 4 sub-blocks respectively. The feature code of the current node is composed of its position 01 in the parent block, the node number 00 of the child block and the dominant attribute distribution template 1001, namely 01001001 (hexadecimal 49).
For 2 stored in the above case12The gray level bright level image is explained with reference to fig. 7(b), in which the current image block is located at the lower left corner (binary 10 position) of the parent image block, and is composed of 4 monotone sub-blocks with different values, and the only dominant attribute sub-block is the 01 sub-block. The current node attribute field stores the packed value (16-bit binary number) of the first 4-bit binary number of the 4 monotonic sub-block attribute values, and the pointer-sub-block attribute association stores the last 8-bit of the 4 sub-block attribute values respectively. The feature code of the current node consists of its position 10 in the parent block, the node number 00 of the child block and the dominance attribute distribution template 0001, namely 10000001 (hexadecimal 81).
If the current image block only contains the non-monotonic and dominant sub-block images, a sub-block node table with corresponding length is created according to the number of the non-monotonic sub-block images, and characteristic assignment processing is carried out on the corresponding nodes of each non-monotonic sub-block according to the property of the non-monotonic sub-block. With reference to fig. 8, the current block in the diagram is located at the lower right corner (binary 11 position) of the parent block, and is composed of 2 non-monotonic sub-blocks and 2 dominant monotonic sub-blocks, the established node table stores the nodes corresponding to the 2 non-monotonic sub-blocks, the number of the sub-blocks is 2 (binary 10), the number of the 2 dominant monotonic sub-block distribution templates is 9 (binary 1001), and the feature code of the current node is composed of the position 11 of the current node in the parent block, the number of the sub-block nodes is 10, and the dominant attribute distribution template is 11101001 (hexadecimal E9);
if the current image block contains non-monotonic or non-dominant sub-block images, a sub-block node table with corresponding length is created according to the number of the sub-block images, and feature assignment processing is carried out on corresponding nodes of each non-monotonic or non-dominant sub-block. With reference to fig. 9, the current block in the figure is located at the lower right corner (binary 11 position) of the parent block, and is composed of 1 non-monotonic sub-block, 2 dominant monotonic sub-blocks and 1 non-dominant monotonic sub-block, the current node attribute field stores dominant attributes, the established sub-block node table stores nodes corresponding to 1 non-monotonic sub-block and 1 non-dominant monotonic sub-block, so the number of sub-block nodes is 2, i.e., binary 10, 2 dominant monotonic sub-block distribution templates is C (binary 1100), and the feature code of the current node is composed of the position 11 in the parent block, the number of sub-block nodes 10, and the dominant attribute distribution template 1100, i.e., 11101100 (hexadecimal EC). 1 non-dominant monotonic sub-block is located at 00 position of current block, so the feature code of corresponding node is 0F;
The characteristic assignment processing judges the image properties of the sub-blocks, if the sub-blocks are non-monotonic sub-blocks, the characteristic codes of the corresponding nodes are obtained by splicing according to the positions of the characteristic codes in the father block, the number of child nodes and a monotonic dominant sub-block distribution template, and attribute values of the monotonic dominant sub-blocks and the head address-attribute value complex values of the child node tables are given to the corresponding nodes;
the characteristic assignment processing judges whether the sub-block image is a non-dominant monotonic sub-block, the corresponding node is a leaf node, the number of child nodes is 0, the characteristic code of the corresponding node is obtained by splicing according to the position of the corresponding node in a father block and a monotonic dominant sub-block distribution template, the attribute value of the monotonic dominant sub-block is given to the corresponding node, and the child pointer is given with NULL;
and returning the first address of the node table of the sub block of the current block.
In order to verify the performance of the enhanced advantageous quadtree, different types of quadtree structures are adopted for carrying out compression storage experiments aiming at 3 digital images with different scales and complexity so as to compare the performance expression of the compression storage digital images of the different types of quadtree structures.
FIG. 10 shows the compression performance comparison data for a conventional quadtree, a linear quadtree, a dominant quadtree, and an enhanced quadtree, from which the salient performance advantage of the enhanced dominant quadtree can be seen, for storing a 4096 × 4096 digital image, the compression ratio of the enhanced dominant quadtree is 10.58 times that of the conventional quadtree, 2.34 times that of the linear quadtree, and 1.45 times that of the dominant quadtree; for storing one 5528 x 5596 digital image, the compression ratio of the enhanced dominant quadtree is 10.91 times that of the conventional quadtree, 2.33 times that of the linear quadtree and 1.45 times that of the dominant quadtree; for storing a 4184 x 6432 digital image, the compression ratio of the enhanced dominant quadtree is 10.81 times that of the conventional quadtree, 3.18 times that of the linear quadtree, and 1.39 times that of the dominant quadtree.
Claims (15)
1. An enhanced dominant quadtree structure with performance enhancement features embodied in two aspects: firstly, the child node table pointer domain of the dominant quadtree structure is transformed into a monotonic sub-block attribute-child node table pointer complex, the proportion of information nodes in the tree is further improved, the depth of the tree is effectively reduced, and the higher lossless compression ratio of raster image storage compared with the dominant quadtree is realized: secondly, the dominant quadtree structure is expanded to be considered as 28Gray scale sum of luminance 212The gray level and bright level image data is stored in a lossless compression mode, and the applicability of the gray level and bright level image data is enhanced.
2. The enhanced advantageous quadtree structure is particularly characterized by having two data structure forms, one for each of the 2-oriented data structures8Lossless compression storage and 2-oriented of gray level and bright level image data12And lossless compression and storage of the gray level and bright level image data.
3. Face 28The basic data structure features of the gray scale bright level image data storage are as follows: the node type of the tree consists of an unsigned byte (8bit) characteristic information code domain, an unsigned byte dominant monotonic sub-block attribute domain and a 4-byte (32bit) child node table pointer-attribute domain combination.
4. Face 212The basic data structure features of the gray scale bright level image data storage are as follows: the node type of the tree consists of an unsigned byte type characteristic information code domain, an unsigned double-byte short integer (16bit) dominant monotonic sub-block attribute domain and a 4-byte (32bit) child node table pointer-attribute domain combination.
5. The pointer-attribute complex of the enhanced advantageous quadtree structure is used as a pointer field for storing the head address of the node table of the subblock image expanded by the block corresponding to the current node only when the block corresponding to the current node has the non-monotonic subblock image.
6. For storage 28In the enhanced advantageous quadtree structure of the gray level and brightness level image data, if the block corresponding to the current node is composed of 4 monotone sub-block images with incompletely same values, the current node is not expanded downwards to become a leaf node, and 4 byte spaces of a pointer-attribute complex of the leaf node are used as storage units to respectively store gray values or attribute values of the 4 monotone sub-block images.
7. For storage 212In the enhanced advantageous quadtree structure of the gray level and brightness level image data, if a current node corresponding to an image block is composed of 4 monotone sub-block images which are not completely identical in value, the current node does not expand downwards to become a leaf node, 4 byte spaces of a pointer-attribute complex of the current node respectively store the last 8-bit binary data of 4 monotone sub-block gray values, and the first 4-bit binary data of the 4 monotone sub-block gray values are sequentially spliced into 16-bit binary data to be stored in an attribute domain of the node.
8. In the strong advantageous quadtree structure, when the current node corresponding to the block does not contain a monotonic sub-block or consists of only non-monotonic sub-blocks, the attribute domain of the current node is empty, the non-monotonic sub-blocks within the bounded area are represented by the extended child nodes, and the pointer-attribute associations of the nodes store the first addresses pointing to the child node tables.
9. In the strong dominance quadtree structure, when the current node corresponding to the block contains both monotone and non-monotone sub-blocks, the attribute domain of the current node stores the gray value or attribute value of the dominance monotone sub-block, the number of which is most, the non-monotone sub-block and the non-dominance monotone sub-block are represented by the extended child node, and the pointer-attribute association of the node stores the head address pointing to the child node table.
10. The definition of the enhanced dominant quadtree node characteristic information code is expanded into: when the node number node in the feature code field is 0, if 0 < template < F (16 system), then 4 bytes of the current node pointer-attribute complex are respectively used for storing the low 8-bit binary gray value of 4 monotonic sub-blocks, and for 2 pairs12The gray scale image is characterized in that the high 4 bits of binary gray values of 4 monotonic sub-blocks are spliced into 16-bit binary data in sequence and stored in the attribute domain of the node.
11. The method for constructing the enhanced superior quadtree comprises the steps of firstly establishing a root node corresponding to the whole normalized image or grid data, and adopting a top-down recursive quadtree mechanism and a bottom-up four-neighborhood sub-block characteristic analysis and node construction regression link mechanism to establish a complete enhanced superior quadtree.
12. The enhanced dominant quadtree construction method assembles the feature code of the current node according to the analysis of the feature information parameters of 4 sub-blocks of a corresponding image block of the current node, assigns a dominant attribute value or a high 4-bit assembly value of a binary gray value of a monotonic sub-block, and assigns a low 8-bit of the created sub-block node table address or the binary gray value of the 4 monotonic sub-blocks to a pointer-attribute association of the current node.
13. The enhanced advantageous quadtree digital image/raster data storage reading method adopts a quadtree-based quick search mechanism, judges whether a current image block is completely monotonous, is completely composed of monotonous subblocks with incompletely same values, is composed of non-monotonous subblocks and monotonous subblocks, or is only composed of non-monotonous subblocks by analyzing a node feature code corresponding to the current image block, and adopts the following corresponding access mechanism.
14. And directly returning the gray value in the attribute domain of the current image block in the situation that the current image block in which the read image element is positioned is completely monotonous. For the situation that the current image block where the read image element is located is composed of all the monotone sub-blocks, the gray value stored in the corresponding byte in the pointer-attribute association is read according to the sub-block number where the specified read image element is located, and for 2 12And in the gray scale image, the single-byte gray scale value is also required to be spliced with a corresponding high 4-bit binary gray scale value in the attribute domain to form a 12-bit binary gray scale value as a read gray scale value.
15. And further judging whether the read pixel is in the dominant monotonic sub-block range or not under the condition that the current image block where the read pixel is located is composed of the non-monotonic sub-block and the monotonic sub-block, directly reading the gray value of the current node attribute domain for the pixel in the dominant monotonic sub-block range, and continuously searching for the child node corresponding to the sub-block where the pixel is located for the pixel in the non-monotonic sub-block range.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910357775.7A CN111862247A (en) | 2019-04-29 | 2019-04-29 | Enhanced dominant quadtree spatial data structure and construction method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910357775.7A CN111862247A (en) | 2019-04-29 | 2019-04-29 | Enhanced dominant quadtree spatial data structure and construction method thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111862247A true CN111862247A (en) | 2020-10-30 |
Family
ID=72966498
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910357775.7A Pending CN111862247A (en) | 2019-04-29 | 2019-04-29 | Enhanced dominant quadtree spatial data structure and construction method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111862247A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0518464A2 (en) * | 1991-06-14 | 1992-12-16 | Tektronix Inc. | Adaptive spatio-temporal compression/decompression of video image signals |
| DE19503571A1 (en) * | 1994-09-27 | 1996-03-28 | Samsung Electronics Co Ltd | Video data compression method |
| EP1009160A1 (en) * | 1998-12-09 | 2000-06-14 | Sharp Kabushiki Kaisha | A quad-tree embedded image compression and decompression method and apparatus |
-
2019
- 2019-04-29 CN CN201910357775.7A patent/CN111862247A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0518464A2 (en) * | 1991-06-14 | 1992-12-16 | Tektronix Inc. | Adaptive spatio-temporal compression/decompression of video image signals |
| DE19503571A1 (en) * | 1994-09-27 | 1996-03-28 | Samsung Electronics Co Ltd | Video data compression method |
| EP1009160A1 (en) * | 1998-12-09 | 2000-06-14 | Sharp Kabushiki Kaisha | A quad-tree embedded image compression and decompression method and apparatus |
Non-Patent Citations (3)
| Title |
|---|
| 张志强等: "SESQ系统的一种查询优化策略", 《计算机研究与发展》 * |
| 沈定涛等: "一种顾及属性的游程编码"交"运算方法与实验", 《地理与地理信息科学》 * |
| 谢顺平等: "一种基于优势属性存储的四叉树结构及其构建算法", 《武汉大学学报·信息科学版》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112287138B (en) | Organization scheduling method, device and equipment of city information model | |
| WO2022042539A1 (en) | Spatial order-based point cloud layering method, point cloud prediction method and device | |
| CN105144141B (en) | For using the system and method apart from relevance hashing to media database addressing | |
| CN101719154B (en) | Grid structure-based spatial index establishing method and grid structure-based spatial index establishing system | |
| CN104378644A (en) | Fixed-width variable-length pixel sample value string matching strengthened image compression method and device | |
| KR20020031015A (en) | Non-linear quantization and similarity matching methods for edge histogram bins | |
| CN107798054A (en) | A kind of range query method and device based on Trie | |
| CN110457317B (en) | Hilbert curve coding and decoding method based on state view | |
| CN115606188A (en) | Point cloud encoding and decoding method, encoder, decoder and storage medium | |
| CN108259911A (en) | A kind of OLED screen Demura lossless date-compress, decompression method | |
| WO2022068745A1 (en) | Data processing method and device | |
| JP2001211081A (en) | Digital array re-indexing method and digital data compressor | |
| CN106204719A (en) | Magnanimity model real-time scheduling method in three-dimensional scenic based on two-dimensional neighbourhood retrieval | |
| Pajarola et al. | An image compression method for spatial search | |
| JP2023510081A (en) | Attribute information prediction method, encoder, decoder and storage medium | |
| CN111862247A (en) | Enhanced dominant quadtree spatial data structure and construction method thereof | |
| Mohamed et al. | Speeding up fractal image compression by genetic algorithms | |
| Li et al. | MGeohash: Trajectory data index method based on historical data pre-partitioning | |
| CN113076334A (en) | Data query method, index generation device and electronic equipment | |
| Jackson et al. | Parallel pipelined fractal image compression using quadtree recomposition | |
| Giudice et al. | A fast palette reordering technique based on GPU-optimized genetic algorithms | |
| Koc et al. | Application of pseudo-distance to lossless coding of color-mapped images | |
| CN117312471B (en) | Sea-land attribute judging method and device for massive position points | |
| Unnikrishnan et al. | Threaded linear hierarchical quadtrees for computation of geometric properties of binary images | |
| CN1117334C (en) | Method for reducing memory capacity and creation of database |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20201030 |
|
| WD01 | Invention patent application deemed withdrawn after publication |