WO2020198997A1

WO2020198997A1 - Connected domain processing method, data processing device and computer readable storage medium

Info

Publication number: WO2020198997A1
Application number: PCT/CN2019/080583
Authority: WO
Inventors: 李泽祥; 高明明; 杨康; 李涛
Original assignee: 深圳市大疆创新科技有限公司
Priority date: 2019-03-29
Filing date: 2019-03-29
Publication date: 2020-10-08
Also published as: CN111213178A

Abstract

A connected domain processing method, a data processing device and a computer readable storage medium, wherein the method comprises: obtaining an image to be analyzed, performing a connected domain analysis on the current pixel points to be analyzed in the image to be analyzed, and recording a target connected domain label for the pixel points to be analyzed in a label map of the image to be analyzed according to an analysis result; updating connected domain statistical information of the image to be analyzed according to the target connected domain label; if the target connected domain label is a root node label, recording first-type statistical data for the target connected domain label in the connected domain statistical information, wherein the first-type statistical data is number data; and if the target connected domain label is an intermediate node label, recording second-type statistical data for the target connected domain label in the connected domain statistical information, wherein the second-type statistical data is pointer data. The present invention not only can effectively reduce the number of read requests for DDR, but also can realize the multiplexing of the statistical data, thereby effectively saving the storage space and computing resources, and enlarging the application scenes of the connected domain processing.

Description

Connected domain processing method, data processing equipment and computer readable storage medium

Technical field

The present invention relates to the field of computer technology, in particular to a connected domain processing method, data processing equipment and computer-readable storage medium.

Background technique

Connected domain processing is an important part of image recognition technology. For example, connected domain processing can be performed on images to realize license plate recognition, text recognition, caption recognition, vehicle tracking, and pedestrian intrusion detection. At present, the connected domain processing method widely used in the industry is the connected domain processing method based on "seed diffusion". However, when the connected domain processing method based on "seed diffusion" is adopted, once inter-bank processing occurs, an off-chip double-rate synchronous dynamic random access memory (Double Data Rate, DDR) read request needs to be initiated. Since frequent access to DDR will take up a lot of bandwidth resources, in actual use, the connected domain processing is usually performed on smaller images (such as 640*360 pixels) based on "seed diffusion". For larger images, downsampling is required before connected domain processing can be performed based on "seed diffusion", for example, an image with a size of 1280*720pixel is downsampled to 640*360pixel for processing. However, the down-sampling operation will reduce the accuracy of the image and affect the accuracy of the processing result. It can be seen that the connected domain processing method based on "seed diffusion" requires a large number of DDR read requests, and the application scenarios are very limited.

Summary of the invention

The embodiment of the present invention discloses a connected domain processing method, data processing equipment, and computer-readable storage medium, which can not only effectively reduce the number of DDR read requests, but also realize the multiplexing of statistical data, effectively saving storage space and computing resources, Expand the application scenarios of connected domain processing.

The first aspect of the embodiments of the present invention discloses a connected domain processing method, and the method includes:

Acquiring an image to be analyzed, performing a connected domain analysis on the current pixel to be analyzed in the image to be analyzed, and recording a target connected domain label for the pixel to be analyzed in the label map of the image to be analyzed according to the analysis result;

Updating the connected domain statistical information of the image to be analyzed according to the target connected domain label;

Wherein, when the connected component statistical information is updated, if the target connected component label is a root node label, then the first type of statistical data is recorded for the target connected component label in the connected component statistical information, so The first type of statistical data is number data, and the number data is used to indicate the number of pixels marked as the target connected domain label and the marked connected domain label pointing to the target connected domain label; if the target is connected If the domain label is an intermediate node label, then the second type of statistical data is recorded for the target connected domain label in the connected domain statistical information, the second type of statistical data is pointer data, and the pointer data is used to indicate the The label of the root node pointed to by the target connected domain label.

The second aspect of the embodiments of the present invention discloses a data processing device, which includes a memory and a processor,

The memory is used to store program instructions;

The processor is configured to execute program instructions stored in the memory, and when the program instructions are executed, the processor is configured to:

A third aspect of the embodiments of the present invention discloses a computer-readable storage medium in which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described in the first aspect are implemented .

In the embodiment of the present invention, the connected component analysis is performed on the pixel to be analyzed first, and the target connected component label is recorded in the label map for the pixel to be analyzed according to the analysis result; then the connected component statistical information is updated according to the target connected component label; where, if If the target connected domain label is the root node label, the first type of statistical data is recorded in the connected domain statistical information, and if the target connected domain label is an intermediate node label, the second type of statistical data is recorded in the connected domain statistical information. By adopting the above method, not only can the number of DDR read requests be effectively reduced, but also the multiplexing of statistical data can be realized, storage space and computing resources can be effectively saved, and the application scenarios of connected domain processing can be expanded.

Description of the drawings

In order to explain the technical solutions in the embodiments of the present invention more clearly, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative labor, other drawings can be obtained from these drawings.

Fig. 1 is a schematic diagram of a connected domain disclosed in the first embodiment of the present invention;

2 is a schematic diagram of a label map disclosed in an embodiment of the present invention;

FIG. 3 is a schematic diagram of a polytree disclosed in an embodiment of the present invention;

4 is a schematic flowchart of a connected domain processing method disclosed in the first embodiment of the present invention;

Fig. 5 is a schematic structural diagram of connected domain statistical information disclosed in the first embodiment of the present invention;

Fig. 6 is a schematic diagram of the result after the connected domain analysis of the pixels in the first row of Fig. 1 is completed;

Fig. 7 is a schematic diagram of the result after the connected domain analysis of the second row of pixels in Fig. 1 is completed;

8 is a schematic flowchart of a connected domain processing method disclosed in the second embodiment of the present invention;

9 is a schematic diagram of a connected domain disclosed in the second embodiment of the present invention;

FIG. 10 is a schematic diagram of row label information disclosed in an embodiment of the present invention;

11 is a schematic structural diagram of connected domain statistical information disclosed in the second embodiment of the present invention;

12 is a schematic diagram of a connected domain disclosed in the third embodiment of the present invention;

FIG. 13 is a schematic structural diagram of connected domain statistical information disclosed in the third embodiment of the present invention;

14 is a schematic structural diagram of connected domain statistical information disclosed in the fourth embodiment of the present invention;

15 is a schematic structural diagram of connected domain statistical information disclosed in the fifth embodiment of the present invention;

Fig. 16 is a schematic structural diagram of a data processing device disclosed in an embodiment of the present invention.

detailed description

The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention.

The connected domain processing method in the embodiment of the present invention is implemented based on the connected domain analysis method of "pixel labeling". To better understand this solution, the following first describes the connected domain analysis method based on "pixel labeling". The connected domain analysis method based on "pixel marking" can directly operate on the original image, mainly to reduce the number of DDR read requests during the connected domain analysis process. Taking four-connectivity analysis as an example, this method can only judge whether the pixel to be analyzed is The pixel points adjacent to the "up" direction position and/or the "left" direction position of its position are connected. The data needed in the connected domain analysis process includes: 1. Connected domain statistical data, which can be recorded as result, which can be used to record the number of pixels marked as connected domain labels. Generally, the maximum value of the connected domain label is determined according to experience. 2. The pointer data whose depth is consistent with the maximum value of the connected domain label can be recorded as pointer_list, which is used to form the connected domain into a polytree. 3. The label map of the image to be analyzed can be recorded as intlabel_list. The depth of the label map is consistent with the size of the image to be analyzed, and is used to store the connected domain label of each pixel in the image to be analyzed.

The following is an example of a connected domain analysis method based on "pixel labeling", specifically taking four connected analysis as an example. Please also refer to FIG. 1, which is a schematic diagram of a connected domain provided by the first embodiment of the present invention. As shown in Figure 1, Figure 1 shows the image to be analyzed, and each grid in the figure represents a pixel of the image to be analyzed. According to the position of the grid in the image to be analyzed, or according to the position of the pixel indicated by the grid in the image to be analyzed, mark information is set for each grid. The marking information is used to indicate the number of rows and columns of the pixels represented by the grid or the grid in the image to be analyzed, that is, the coordinate information; the first identifier in the marking information is used to indicate the number of rows, the second in the marking information Bit identification is used to indicate the number of columns. For example, the marking information 22 indicates that the position of the grid or the pixel represented by the grid is the second row and the second column in the image to be analyzed; for example, the marking information 1b indicates that the position of the pixel represented by the grid or the grid is in the image to be analyzed. The first row and the eleventh column.

The shaded grid in the figure represents the effective pixels in the image to be analyzed, and the unshaded grid (or white grid) represents the invalid pixel in the image to be analyzed; the multiple grids filled with shadow constitute the majority of the image to be analyzed. Connected domains. In the process of analyzing the image to be analyzed based on the connected domain analysis method of "pixel labeling", the effective pixels in the image to be analyzed are not connected with the invalid pixels, and the connected domain labels of the invalid pixels in the image to be analyzed can be directly Determine as an invalid connected domain label (for example, a value of 0); determine the valid pixel in the image to be analyzed as the pixel to be analyzed. Then determine whether the pixel to be analyzed is connected to the pixel that has been assigned a connected domain label. If the pixel to be analyzed is connected to the pixel that has been assigned a connected domain label, then from the connected domain label of the pixel connected to the pixel to be analyzed , To determine the target connected domain label of the pixel to be analyzed; if the pixel to be analyzed is not connected to the pixel that has been assigned a connected domain label, a new connected domain label is assigned to the pixel to be analyzed, and the pixel to be analyzed The given connected component label is determined as the target connected component label of the pixel to be analyzed.

Specifically, first initialize the connected domain statistical data result, pointer data pointer_list, and label map intlabel_list to 0. Assuming that the value of the largest supported connected component label is 10, it means that the available connected component labels are values 1-10. In the process of analyzing the connected domains of the image to be analyzed shown in Fig. 1, the sequence is from left to right and top to bottom. It can be seen from Fig. 1 that the first effective pixel in the image to be analyzed is the pixel indicated by the shaded filled grid with the label information 12, which is hereinafter referred to as pixel 12, and the following description is analogous to this. Since there is no pixel in the "up" direction where the pixel 12 is located in the image to be analyzed, the pixel 11 at the "left" position of the pixel 12 is an invalid pixel and is not connected with the pixel 12; At this time, a new connected component label needs to be assigned to the pixel 12, and the connected component label 1 can be determined as the target connected component label of the pixel 12. Then the connected component statistical data result[1] corresponding to the connected component label 1 is set to 1, and the pointer data point_list[1] corresponding to the connected component label 1 is set to 1, indicating that the connected component label 1 is a root node label. Further, please also refer to FIG. 2, which is a schematic diagram of a label map provided by an embodiment of the present invention, specifically the label map of the image to be analyzed shown in FIG. 1. As shown in Figure 2, after the target connected domain label of the pixel 12 is determined to be 1, the connected domain label 1 can be recorded for the pixel 12 in the label map of the image to be analyzed as shown in Figure 2, that is, the label image The label_list[1][2] in the first row and second column position in the label is 1.

Further, when the connected component analysis is performed on the pixels 16 and 19, by analogy, it can be determined that the target connected component labels of the pixels 16 and 19 are 2 and 3, respectively. When analyzing the connected domain of the pixel point 1a, because the pixel point 1a is connected to the pixel point 19 adjacent to the "left" direction of its location, and the target connected domain label of the pixel point 19 is 3; then the pixel point 1a The target connected domain label is also determined to be 3. And adjust the connected component statistical data result[3] corresponding to connected component label 3 to 2 (that is, two pixels of 19 and 1a). When analyzing the connected component of the pixel 21, it can be determined that the target connected component label of the pixel 21 is 4. When the connected domain analysis is performed on the pixel point 22, the pixel point 22 is connected with the pixel point 12 adjacent to the "up" direction of its location, and connected with the pixel point 21 adjacent to the "left" direction of its location; The target connected domain label of pixel 12 is 1, and the target connected domain label of pixel 21 is 4; it indicates that pixel 12 and pixel 21 belong to the same connected domain. Correspondingly, connected

domain labels

1 and 4 also belong to the same connected domain. Connected domain. At this time, the target connected component label of the pixel 22 is determined to be 1 with a smaller label value, and the two connected components corresponding to the connected component label 1 and the connected component label 4 need to be merged. The pointer data point_list[4] corresponding to connected component label 4 can be adjusted to 1, indicating that connected component label 4 is an intermediate node label and points to root node label 1. At the same time, the connected component statistical data corresponding to connected component label 1 Adjust result[1] to 3 (that is, 12, 21, and 22 pixels). Among them, the connected domain statistical data result[4] corresponding to the connected domain label 4 can be adjusted to 1 at the same time, or adjusted to the value 3 recorded in result[1]. It should be noted that the processing procedures not mentioned in the above connected domain analysis process can refer to related descriptions in the connected domain analysis process for the pixel 12, and will not be repeated here.

The connected domain analysis of the image to be analyzed is completed by the above processing method, and the label map of the image to be analyzed as shown in FIG. 2 can be obtained, and the polytree shown in FIG. 3 can be obtained according to the recorded pointer data point_list. Based on the label map, the connected domain label of each pixel in the image to be analyzed can be determined, and the relationship between the connected domain labels used to mark the pixel to be analyzed can be determined based on the multi-branch tree. As shown in Figure 3, it can be seen that connected domain labels 1, 6, and 8 are root node labels, connected domain labels 2, 3, 4, 5, 7, and 9 all point to root node label 1, and connected domain label 10 points to root node label 8. In addition, according to the recorded connected component statistical data result or label map, the number of pixels marked as connected component label 1 and the labeled connected component label pointing to connected component label 1 is 36, and the number of pixels marked as connected component label 8 and labeled connected component The number of pixels where the connected domain label points to the connected domain label 8 is 5, and the number of pixels marked as connected domain label 6 is 1. Based on the data obtained by the above-mentioned connected domain processing, the image to be analyzed can be processed continuously, for example, pixels in the connected domain whose number of connected pixels in the image to be analyzed are less than a preset value are filtered.

It should be noted that, since the two pixels adjacent to the "up" and "left" positions of the pixel to be analyzed can be obtained from the continuous encoded data stream of DDR, the above is adopted To complete the marking of an entire image, only one DDR read request needs to be initiated. It can be seen that the connected domain analysis method based on "pixel tagging" can effectively reduce the problem of too many DDR read requests and a large amount of DDR bandwidth based on the connected domain processing method based on "seed diffusion"; thus, the connected domain analysis based on "pixel tagging" The method can also perform connected domain processing on larger images. However, when the connected domain analysis method based on "pixel labeling" is used, when the current pixel to be analyzed is not connected to the pixel that has been assigned a label, a new label needs to be assigned to the current pixel to be analyzed; As a result, an image needs to use a large number of connected domain labels, that is, the maximum value of connected domain labels required by an image is very large. For example, the number of connected domain tags that need to be used when processing an image with a size of 1280*720pixel reaches more than 33,000. At the same time, each connected domain label needs to be assigned a pointer data or connected domain statistical data; both pointer data and connected domain statistical data need to be continuously read and written when scanning images, and stored in the on-chip memory; therefore, it will take up a lot of storage Space, to a certain extent, limits the application scenarios of connected domain analysis based on "pixel labeling".

In order to solve the above problems, the embodiments of the present invention provide a connected domain processing method, which can not only effectively reduce the number of DDR read requests, but also realize the multiplexing of statistical data, effectively save storage space and computing resources, and expand the connected domain processing Application scenarios. Detailed descriptions are given below.

Refer to FIG. 4, which is a schematic flowchart of a connected domain processing method according to the first embodiment of the present invention. The connected domain processing method described in the embodiment of the present invention may include:

S401: The data processing device acquires an image to be analyzed, and performs connected domain analysis on the current pixel to be analyzed in the image to be analyzed.

In the embodiment of the present invention, the data processing device first obtains an image to be analyzed, and the image to be analyzed includes a plurality of pixels, and the plurality of pixels includes valid pixels and invalid pixels. The effective pixels and invalid pixels in the image to be analyzed are not connected. In an embodiment, the image to be analyzed may be any one of a plurality of depth images collected by a multi-lens camera taking a picture of the subject at a time. If the depth information between the pixels in the same position of the image to be analyzed and the contrast depth image is within the error range, the pixel at that position in the image to be analyzed is determined as the effective pixel; otherwise, the pixel in the image to be analyzed is determined The pixel at this position is determined to be an invalid pixel; where the contrast depth image is the depth image excluding the image to be analyzed among the multiple depth images. In another embodiment, the image to be analyzed may be any one of two depth images collected by a binocular camera at a time by taking pictures of the subject. First, use the image to be analyzed as the reference image and the contrast depth image as the matching image to perform the depth information matching operation to obtain the first matching result; the first matching result is used to indicate the pixel points between the image to be analyzed and the contrast depth image at the same position The first matching value. Then use the contrast depth image as the reference image and the image to be analyzed as the matching image to perform the depth information matching operation to obtain the second matching result; the second matching result is used to indicate the pixel points between the image to be analyzed and the contrast depth image at the same position The second matching value. If the first matching value and the second matching value between the pixels at the same position of the image to be analyzed and the contrast depth image are the same or within the error range, the pixel at that position in the image to be analyzed is determined as an effective pixel Point; otherwise, the pixel at that position in the image to be analyzed is determined as an invalid pixel. Wherein, the contrast depth image is the other one of the two depth images except the image to be analyzed.

In an embodiment, the data processing device may first directly determine the connected domain label of the invalid pixel in the image to be analyzed as the invalid connected domain label, for example, the value 0; The effective pixels in the analysis image are determined as the pixels to be analyzed, and then only the pixels to be analyzed in the image to be analyzed are subjected to connected domain analysis to reduce the amount of data processing, and save software and hardware resources and processing time. Further, the data processing device performs connected domain analysis on the current pixel to be analyzed in the image to be analyzed to obtain an analysis result; the analysis result includes the target connected domain label of the pixel to be analyzed. The data processing equipment first determines the adjacent pixels adjacent to the pixel to be analyzed from the image to be analyzed, and then performs connected domain analysis on the pixel to be analyzed based on the adjacent pixel to determine the target connected domain of the pixel to be analyzed label.

In an embodiment, the adjacent pixels adjacent to the pixel to be analyzed include: the first pixel adjacent to the pixel to be analyzed in the same column in the upper direction, and/or, the pixel to be analyzed The second pixel point adjacent to the left in the same row. In an embodiment, the data processing device performs a connected domain analysis on the pixel to be analyzed based on the first pixel and the second pixel, and determines whether the pixel to be analyzed is connected to the first pixel and the second pixel. If it is determined that the pixel to be analyzed is connected to the first pixel, the connected domain label of the first pixel is used as the target connected domain label of the pixel to be analyzed. If it is determined that the pixel to be analyzed is connected to the second pixel, the connected domain label of the second pixel is used as the target connected domain label of the pixel to be analyzed. If it is determined that the pixel to be analyzed is connected to the first pixel and the second pixel, the connected domain label with the smallest value among the connected domain labels of the first pixel and the connected domain labels of the second pixel is used as the connected domain label to be analyzed The target connected domain label of the pixel. If it is determined that the pixel to be analyzed is not connected to the first pixel and the second pixel, then an unused connected component label among the preset connected component labels is used as the target connected component label of the pixel to be analyzed .

Among them, the way to determine whether the pixel to be analyzed is connected to the first pixel and the second pixel may be: first obtain the value corresponding to the pixel to be analyzed, the first pixel and the second pixel, and the value may be depth Value; then compare the value corresponding to the pixel to be analyzed with the value corresponding to the first pixel to obtain the first comparison result; compare the value corresponding to the pixel to be analyzed with the value corresponding to the second pixel , Get the second comparison result. If the first comparison result indicates that the difference between the value corresponding to the pixel to be analyzed and the value corresponding to the first pixel is less than or equal to the preset threshold, it is determined that the pixel to be analyzed is connected to the first pixel; otherwise, It is determined that the pixel to be analyzed is not connected to the first pixel. Similarly, if the second comparison result indicates that the difference between the value corresponding to the pixel to be analyzed and the value corresponding to the second pixel is less than or equal to the preset threshold, it is determined that the pixel to be analyzed is connected to the second pixel ; Otherwise, it is determined that the pixel to be analyzed is not connected with the second pixel.

In one embodiment, the data processing device performs connected domain analysis on the pixel to be analyzed based on the first pixel to determine whether the pixel to be analyzed is connected to the first pixel; if it is determined that the pixel to be analyzed is connected to the first pixel, Then the connected domain label of the first pixel is used as the target connected domain label of the pixel to be analyzed. If it is determined that the pixel to be analyzed is not connected to the first pixel, an unused connected domain label among the preset connected domain labels is used as the target connected domain label of the pixel to be analyzed. In one embodiment, the data processing device performs connected domain analysis on the pixel to be analyzed based on the second pixel to determine whether the pixel to be analyzed is connected to the second pixel; if it is determined that the pixel to be analyzed is connected to the second pixel, Then, the connected domain label of the second pixel is used as the target connected domain label of the pixel to be analyzed. If it is determined that the pixel to be analyzed is not connected to the second pixel, an unused connected domain label among the preset connected domain labels is used as the target connected domain label of the pixel to be analyzed.

In another embodiment, the adjacent pixels adjacent to the pixel to be analyzed include: the first pixel adjacent to the pixel to be analyzed in the same column and in the upper direction, and in the same row as the pixel to be analyzed One or more of the second pixel point adjacent to the left direction, the third pixel point adjacent to the pixel to be analyzed in the upper left direction, and the fourth pixel point adjacent to the pixel to be analyzed in the upper right direction . Among them, the third pixel is the pixel that is adjacent to the left in the same row as the first pixel, and the fourth pixel is the pixel that is adjacent to the right in the same row as the first pixel. point. In an embodiment, the data processing device performs a connected domain analysis on the pixel to be analyzed according to the first pixel, the second pixel, the third pixel, and the fourth pixel, and determines whether the pixel to be analyzed is the same as the first pixel, The second pixel point, the third pixel point and the fourth pixel point are connected. If it is determined that the pixel to be analyzed is connected to one of the first pixel, the second pixel, the third pixel, and the fourth pixel, the connected domain label of the pixel connected to the pixel to be analyzed is taken as The target connected domain label of the pixel to be analyzed. If it is determined that the pixel to be analyzed is connected to at least two of the first pixel, the second pixel, the third pixel, and the fourth pixel, then at least two pixels connected to the pixel to be analyzed will be Among the connected component labels, the connected component label with the smallest label value is used as the target connected component label of the pixel to be analyzed. If it is determined that the pixel to be analyzed is not connected with the first pixel, the second pixel, the third pixel, and the fourth pixel, then the unused connected domain label among the preset connected domain labels As the target connected domain label of the pixel to be analyzed.

In one embodiment, this solution also modifies the scanning process in the connected domain analysis method based on "pixel labeling", and adds the line buffer operation, that is, to record line label information and line label information. It is used to record the connected domain label of the currently analyzed pixel. The depth of the line label information is the same as the number of pixels in a line of the image to be analyzed. Wherein, the connected domain label recorded by the row label information includes: the connected domain label of the first pixel adjacent to the upper direction in the same column as the pixel to be analyzed, and the pixel located in the left direction of the same row of the pixel to be analyzed The connected component label of and the connected component label of the pixel in the right direction in the same row of the first pixel. In the process of performing connected domain analysis on the current pixel to be analyzed in the image to be analyzed, if you need to obtain the connected domain label of the pixel adjacent to the pixel to be analyzed, you can quickly obtain it directly from the row label information. Compared with the way of searching the linked list, it can effectively speed up the search and improve the efficiency of connected domain analysis. Further, after the analysis of the connected domain of the pixel to be analyzed is completed, the row label information is updated according to the determined target connected domain label of the pixel to be analyzed. Specifically, the recorded connected domain label of the first pixel can be adjusted to the target connected domain label of the pixel to be analyzed in the row label information. Among them, the row label information can be stored in on-chip registers.

S402: The data processing device records a target connected domain label for the pixel to be analyzed in the label map of the image to be analyzed according to the analysis result.

In the embodiment of the present invention, the label map of the image to be analyzed is used to record the target connected domain label of each pixel in the image to be analyzed. After the data processing device performs connected domain analysis on the current pixel to be analyzed in the image to be analyzed to obtain the analysis result, it records the target connected domain label indicated by the analysis result for the pixel to be analyzed in the label map of the image to be analyzed. In an embodiment, the size of the label map is consistent with the size of the image to be analyzed, and the position of the target connected domain label in the label map is the same as the position of the pixel to be analyzed in the image to be analyzed. The label image is stored in the DDR of the data processing device.

S203. The data processing device updates the connected-domain statistical information of the image to be analyzed according to the target connected-domain label; wherein, when the connected-domain statistical information is updated, if the target connected-domain label is the root node Tag, the first type of statistical data is recorded in the connected domain statistical information for the target connected domain tag, the first type of statistical data is number data, and the number data is used to indicate that the tag is the target connected domain The label and the number of pixels where the labeled connected component label points to the target connected component label; if the target connected component label is an intermediate node label, it is recorded as the target connected component label in the connected component statistical information The second type of statistical data, the second type of statistical data is pointer data, and the pointer data is used to indicate the root node label to which the target connected domain label points.

In the embodiment of the present invention, the connected domain statistical information includes a first identification bit and a data bit, the first identification bit is used to indicate the type of statistical data recorded by the data bit, and the data bit is used to indicate the statistical data recorded as the target connected domain label; When the first identification bit is the first identification, the data bit is the target connected domain label and records the first type of statistical data. When the first identification bit is the second identification, the data bit is the target connected domain label and records the second Class statistics. In an embodiment, the first identifier may be the identifier "0", and the second identifier may be the identifier "1". The first identification bit may be the highest bit of the statistical information of the connected domain, and may also be the lowest bit of the statistical information of the connected domain, which is not limited in the embodiment of the present invention. The connected domain statistical information can be stored in the on-chip random access memory (Random Access Memory, RAM), or a statistical data multiplexing memory can be specially set up to store the connected domain statistical information.

The connected domain processing method in the embodiment of the present invention multiplexes the number data and pointer data on the basis of the connected domain analysis method based on "pixel marking", and only needs to add a 1-bit identification bit to distinguish the connected domain statistical information Whether the statistical data recorded for the target connected domain label is number data or pointer data. Compared with the connected domain analysis method based on "pixel labeling", the on-chip storage space required by the above method is greatly reduced, and the storage space can be reduced by about half. Please also refer to FIG. 5, which is a schematic structural diagram of connected domain statistical information provided by the first embodiment of the present invention. As shown in Figure 5, each square in the figure represents 1 bit, and 11 squares in each row, that is, 11 bits correspond to a connected domain label. For example, the 11-bit data in the first row is connected domain label 1. Connected domain statistics. The first square of each row, or the highest bit, indicates the first identifier of the connected domain statistical information, and is used to indicate the type of statistical data recorded by the data bit of the connected domain statistical information. The second to eleventh squares in each row represent the data bits of the connected domain statistical information, which are used to represent the statistical data recorded by the connected domain label. For example, the first square in the first row records 0, that is, the first identification bit of the connected component statistical information of connected component label 1 is 0, which means that the second to the eleventh square in the first row The grid is connected domain label 1 records the number data, that is, the data bit of connected domain statistical information of connected domain label 1 is connected domain label 1 records the number data; thus indirectly indicates that connected domain label 1 is the root node label. The number data is 26 (11010), which means that the number of pixels marked as connected domain label 1 and the marked connected domain label points to connected domain label 1 is 26. For another example, the first square in the third row records 1, that is, the first identification bit of the connected component statistical information of the connected component label 3 is 1, which means that the second to the 11th square in the first row One square is connected domain label 3, which records the pointer data, that is, the data bit of connected domain statistical information of connected domain label 3 is the connected domain label 3 records pointer data; thus, it also indicates that connected domain label 3 is an intermediate node label. The pointer data is 1 (01), which means that the root node label pointed to by the connected component label 3 is the connected component label 1.

In one embodiment, when the data processing device performs connected analysis of the pixel to be analyzed based on the first pixel and the second pixel, if it is determined that the pixel to be analyzed is not connected to the first pixel and the second pixel , When the unused connected component label among the preset multiple connected component labels is used as the target connected component label of the pixel to be analyzed, the target connected component label is used as the root node label. If it is determined that the pixel to be analyzed is connected to the first pixel and the second pixel, the connected domain label with the smallest value among the connected domain labels of the first pixel and the connected domain labels of the second pixel is used as the connected domain label to be analyzed After the target connected component label of the pixel, the connected component label of the connected component label of the first pixel and the connected component label of the second pixel with the largest label value is determined as the intermediate node label pointing to the target connected component label. Further, the data processing device changes the connected domain label with the largest label value in the connected domain statistical information. Specifically, the first identification bit corresponding to the connected component label with the largest label value in the connected component statistical information is adjusted to the second identifier, and the data bit corresponding to the connected component label with the largest label value in the connected component statistical information is The connected domain label with the largest label value records the second type of statistical data, and the recorded second type of statistical data indicates that the connected domain label with the largest label value points to the target connected domain label.

It should be noted that for the case where adjacent pixels adjacent to the pixel to be analyzed in step S401 include the first pixel, the second pixel, the third pixel, and the fourth pixel, if the pixel to be analyzed is determined The point is connected to at least two of the first pixel, the second pixel, the third pixel, and the fourth pixel. In the connected domain label of the at least two pixels that will be connected to the pixel to be analyzed, After the connected component label with the smallest label value is used as the target connected component label of the pixel to be analyzed, the connected component labels of at least two pixels connected to the pixel to be analyzed, except for the connected component label with the smallest label value The domain labels are all determined as intermediate node labels that point to the target connected domain labels. Further, the data processing device changes the statistical data recorded in the connected domain statistical information for the connected domain labels other than the connected domain label with the smallest label value. For the specific modification method, please refer to the foregoing description and will not be repeated here.

In order to better understand this solution, Figure 1 is still used as an example to describe the connected domain processing method in the embodiment of the present invention. Assuming that the maximum connected domain label supported is 16, and the maximum statistical number is 1024, the data required for this solution includes: 1. The connected domain statistical information that supports the maximum connected domain label can be recorded as record_result_list. The data structure is shown in Figure 5. ; Connected domain statistical information can be used to record statistical data for connected domain labels, including number data or pointer data. Since the maximum supported connected domain label is 16, the maximum number of statistics is 1024, plus 1 bit of the first identification bit, the total storage space required for connected domain statistics is 16*11 bits, which can be stored in the on-chip RAM. 2. The row label information can be recorded as root_pointers; the depth of the row label information is the same as the number of pixels in a line of the image to be analyzed, and the width is the number of bits that support the maximum connected domain label; since the maximum supported connected domain label is 16, The width is 4 bits; the number of pixels in a row of the image to be analyzed in Figure 1 is 12, and the storage space required for the row label information is 12*4 bits in total, which can be stored in the on-chip register. 3. The label map of the image to be analyzed can be recorded as label_list; it is used to store the connected domain label of each pixel in the image to be analyzed. The depth of the label image is consistent with the size of the image to be analyzed, and the width is set according to the maximum connected domain label; the total storage space required for the label image is 12*8*4bits, which can be stored in off-chip DDR.

Specifically, the four-connectivity analysis is used as an example to illustrate. First, initialize all the label map label_list to 0; initialize the first identification bit of the connected domain statistical information record_result_list to 0, initialize the data bit to 0; initialize the row label information root_pointers to all 0. In the process of analyzing the connected component of the image to be analyzed, the connected component analysis is performed on the pixel to be analyzed in Figure 1 in the order from left to right and top to bottom, that is, the pixel indicated by the shadow filled grid. It can be seen from Figure 1 that the first pixel to be analyzed (or effective pixel) of the image to be analyzed is the pixel indicated by the shadow filling grid with the label information of 12. It is referred to as pixel 12 in the following description. And so on. Since there is no pixel in the "up" direction where the pixel 12 is located in the image to be analyzed, the pixel 11 at the "left" position of the pixel 12 is an invalid pixel and is not connected with the pixel 12; At this time, a new connected component label needs to be assigned to the pixel 12, the connected component label 1 can be determined as the target connected component label of the pixel 12, and the connected component label 1 is the root node label.

Please refer to Figure 6 together. Figure 6 is a schematic diagram of the results of the connected domain analysis of the first row of pixels in Figure 1, which specifically includes three parts: label map, row label information, and connected domain statistical information. As shown in FIG. 6, after the target connected domain label of the pixel 12 is determined to be 1, the target connected domain label 1 of the pixel 12 is recorded at the position of the first row and second column in the label_list of the image to be analyzed. Record the target connected domain label 1 for the pixel 12 at the position indicated by the second square in the row label information root_pointers, so that the pixel 22 and the pixel 13 can quickly find the connected domain of the pixel 12 according to the row label information label. Adjust the first identification bit corresponding to connected domain label 1 to 0 at the position indicated by the first square of the connected domain statistical information record_result_list, and the data bit indicated in the first square is connected domain label 1 record number data 1, so the data in the first square is expressed as "0/1". It should be noted that the data structure of the connected component statistical information in Figure 6 is a simplified schematic diagram from the data structure shown in Figure 5. The 16 squares in the connected component statistical information in the figure correspond to connected component labels from left to right. 1 to connected component label 16, the data in each grid indicates that the connected component statistical information is the statistical data recorded by the connected component label. The value on the left of "/" in each piece of data represents the value of the first identification bit, and the value on the right of "/" represents the data recorded by the data bit.

Further, when the connected component analysis is performed on the pixels 16 and 19, by analogy, it can be determined that the target connected component labels of the pixels 16 and 19 are 2 and 3, respectively. When analyzing the connected domain of the pixel point 1a, the pixel point 1a is connected to the pixel point 19 adjacent to the "left" direction of the position of the pixel point 1a, and the pixel point 19 can be quickly found according to the recorded row label information root_pointers The target connected component label of is 3; then the target connected component label of pixel 1a is also determined to be 3. The pixel point 1a and the pixel point 19 have the same information in the label map label and the row label information root_pointers; then at the position indicated by the third square of the connected field statistical information record_result_list, the connected field label 3 corresponds to the first identification position Adjust to 0, and the data bit indicated in the third square is connected domain label 3 record number data 2, so the data in the third square is expressed as "0/2".

When analyzing the connected component of the pixel 21, it can be determined that the target connected component label of the pixel 22 is 4. When the connected domain analysis is performed on the pixel point 22, since the pixel point 22 is connected to the pixel point 12 adjacent to its position in the "up" direction, and is connected to the pixel point 21 adjacent to its position in the "left" direction. Indicates that the pixel 12 and the pixel 21 belong to the same connected domain; in addition, according to the recorded row label information, the target connected domain label of the pixel 12 can be quickly found as 1, and the target connected domain label of the pixel 21 is 4, indicating the connected domain Labels 1 and 4 also belong to the same connected domain. At this time, the target connected component label of the pixel 22 is determined to be 1 with a smaller label value, and the two connected components corresponding to the connected component label 1 and the connected component label 4 need to be merged.

Please refer to Figure 7 together. Figure 7 is a schematic diagram of the results of the connected domain analysis of the second row of pixels in Figure 1, including the label map, row label information and connected domain statistical information. As shown in Fig. 7, after the target connected component label of the pixel 22 is determined to be 1, the position of the second row and second column in the label map label_list of the image to be analyzed is the pixel 22 and its target connected component label 1 is recorded. The position indicated by the second square in the row label information root_pointers is the pixel 22 to record its target connected domain label 1. Adjust the first identification bit corresponding to connected domain label 1 to 0 at the position indicated by the first square of the connected domain statistical information record_result_list, and the data bit indicated in the first square is connected domain label 1 record number data 3. Since the label information in the row of Figure 7 records the data after all the pixels in the second row have been processed, the data in the first square is expressed as "0/4", and the connected domain label is 1 also includes pixels 23 . Adjust the first identification bit corresponding to connected domain label 4 to 1 at the position indicated by the fourth square of the connected domain statistical information record_result_list, and the data bit indicated in the fourth square is the connected domain label 4 record pointer Data 1 indicates that the connected domain label 4 is an intermediate node label and points to the root node label 1, so the data in the fourth square is expressed as "1/1". The above-mentioned processing method is adopted until the analysis of the connected domain of the image to be analyzed is completed, which is not repeated here.

It should be noted that, please refer to the two shadow-filled squares in the row label information in Figure 7. Take the first shadow-filled square from left to right as an example. The shadow-filled square records the pixel 21 The target connected component label, after the connected component analysis of the pixel 21 is completed, it is determined that the target connected component label of the pixel 21 is 4, and the connected component label 4 is the root node label. However, in the process of analyzing the connected area of the pixel 22, the connected area label 4 that was originally the root node label is adjusted to an intermediate node label pointing to the root node label 1. If the connected domain label 4 is directly used as the root node when the connected domain analysis is performed on the pixels in the third row of the image to be analyzed, an error may be caused. Therefore, in an embodiment, in the process of analyzing the connected domain of the pixel to be analyzed in each row of the image to be analyzed, if it is necessary to directly obtain from the row label information root_pointers, the pixel to be analyzed is adjacent to the pixel to be analyzed in the same column. When the connected domain label of the pixel point of the pixel point, the root node label can be judged on the connected domain label of the pixel point in the same column that is adjacent to the pixel to be analyzed in the upper direction obtained from the row label information root_pointers; If the connected component label is the root node label, the target connected component label of the pixel to be analyzed is determined based on the connected component label; if it is determined that the connected component label is not the root node label, search from the connected component statistical information record_result_list The root node label pointed to by the connected component label is derived, and the target connected component label of the pixel to be analyzed is determined based on the root node label pointed to by the connected component label. It has been found through practice that there are not many cases where the connected domain label recorded in the row label information is an intermediate node label, and generally only one or two searches can find the root node label pointed to by the intermediate node label in the row label information. The processing speed of connected domains has little effect; in the process of connected domain analysis, recording row label information and analyzing connected domains based on the row label information can effectively improve the processing speed of connected domains.

Referring to FIG. 8, compared with the method shown in FIG. 4, a connected domain processing method provided in another embodiment further includes the following steps after step S403:

S404: After the multiple preset connected domain labels are used up, the data processing device determines a growing label and a non-growing label from the multiple connected domain labels according to the row label information currently recorded.

In the embodiment of the present invention, after the multiple preset connected component labels are used up, that is, after the multiple preset connected component labels are all allocated to the pixels to be analyzed in the image to be analyzed, the connected component label needs to be checked. Recycle and reuse. A growing label is a connected component label that may continue to extend downward, which has a greater impact on subsequent connected component analysis, and should not support recycling and reuse; a non-growing label is a connected component label that will not continue to extend downward, for subsequent connected component analysis The impact is small and can be recycled and reused. Specifically, the data processing device may determine the growing label and the non-growing label from a plurality of preset connected domain labels according to the row label information currently recorded. In one embodiment, the data processing device determines the first label existing in the row label information of the current record from a plurality of preset connected domain labels, and determines the first label as a growth label; A second label that does not exist in the row label information of the current record is determined among the connected domain labels, and the second label is determined as a non-growing label.

In another embodiment, the data processing device determines the first label existing in the row label information of the current record from a plurality of preset connected domain labels, and determines the first label that does not exist in the row label information of the current record. The second label; then according to the recorded connected domain statistics, from the preset multiple connected domain labels, determine the root node label or the third label of the intermediate node label of the first label; finally remove the second label Tags other than the third tag are determined to be non-growth tags, and the first tag and the third tag are determined to be growing tags. In contrast to the aforementioned direct use of the second label that does not exist in the row label information of the current record as the non-growth label, the method described in this embodiment is to use the second label except the third label as the label. Non-growth label, the non-growth label determined in this way can ensure to a certain extent that it has no effect on subsequent connected domain analysis.

In an embodiment, the connected domain statistical information further includes a second identification bit, the second identification bit is used to record the growth identification of the connected domain label, and when the second identification bit is the third identification, it indicates that the connected domain label is a non-growing label, When the second identification bit is the fourth identification, it indicates that the connected domain label is a growth label. Among them, the third identifier may be the identifier "0", and the fourth identifier may be the identifier "1". The second identification bit can be the highest bit of the statistical information of the connected domain, or the lowest bit of the statistical information of the connected domain; the second identification bit can be adjacent to the first identification bit or not adjacent to the first identification bit . When the second identification position is adjacent to the first identification position, the second identification position may be located adjacent to the left of the first identification position, or may be located adjacent to the right of the first identification position.

S405: The data processing device performs connected domain analysis on pixels in the image to be analyzed that have not completed connected domain analysis according to the non-growth tag.

In the embodiment of the present invention, when the pixel to be analyzed in the image to be analyzed has not completed the connected domain analysis, if it is determined that the pixel to be analyzed is connected to one of its neighboring pixels, it will be connected to the pixel to be analyzed. The connected domain label of the adjacent pixels connected by the dot is used as the target connected domain label of the pixel to be analyzed. If it is determined that the pixel to be analyzed is connected to at least two of its neighboring pixels, then among the connected domain labels of at least two neighboring pixels connected to the pixel to be analyzed, the connected domain label with the smallest label value is used as the to-be-analyzed pixel. Analyze the target connected domain label of the pixel. If it is determined that the pixel to be analyzed is not connected with its neighboring pixels, then an unused connected region label in the non-growth label is used as the target connected region label of the pixel to be analyzed. For the specific processing method, please refer to the previous description, which will not be repeated here. Using the above method, the recovery and reuse of connected domain labels can be realized. The processing of the image to be analyzed can still be completed when the largest supported connected domain label is small; the storage space occupied by the recovered labels remains unchanged, so The processing of the image to be analyzed can be completed with a small storage space, effectively saving storage space.

In order to better understand the recovery and reuse methods of connected domain labels in this solution, the following examples are used for detailed description. Please also refer to FIG. 9, which is a schematic diagram of a connected domain according to the second embodiment of the present invention, which is the result after the analysis is completed according to the connected domain analysis method described above. As shown in Figure 9, Figure 9 shows the image to be analyzed, and each grid in the figure represents a pixel of the image to be analyzed. The shaded grids in the figure represent the effective pixels in the image to be analyzed, and the unshaded grids represent the invalid pixels in the image to be analyzed. The effective pixels and the invalid pixels are not connected; the multiple grids filled with shadow constitute the Analyze multiple connected domains of the image. The mark information in the unshaded filled grid in the figure represents the coordinate information of the corresponding pixel in the image to be analyzed, and the mark information in the shaded grid in the figure represents the coordinate information of the corresponding pixel in the image to be analyzed, and the connected domain Connected domain label determined after analysis. Specifically, the value on the left of "/" indicates the coordinate information of the pixel in the image to be analyzed, and the value on the right of "/" indicates the connected domain label of the pixel.

For the recovery and reuse method of connected domain labels, recovery means to determine non-growth labels, and reuse means to use non-growth labels for subsequent connected-domain analysis; mainly based on a certain segmentation line of the image to be analyzed, if If a certain part of the connected domain in the analysis image does not cross the dividing line, all connected domain labels corresponding to pixels in this part of the connected domain will not appear below the dividing line. Such connected domain labels can be regarded as non-growing labels , And recycle and reuse. Taking the dotted box in Figure 9 as the dividing line as an example, the labels in the connected domains that no longer grow before the line of the dotted box will not appear in the area below the dotted box, for example, the connected domain composed of

connected domain labels

1 and 4 , And the connected domain composed of

connected domain labels

2 and 5 will no longer grow downward, and the connected domain labels 1, 2, 4, and 5 will not appear in the area below the dashed box. The connected domains composed of the connected domain labels in the dashed frame are likely to continue to grow, so the connected domain labels in the dashed frame may appear in the area below the dashed frame, for example, the connected

domain labels

3 and 7 are in the dashed frame. , The connected domain composed of the connected domain label 7 will continue to grow downward, and the connected domain composed of the connected domain label 3 is likely to continue to grow downward, so the connected

domain labels

3 and 7 may both be in the area below the dotted box appear. In summary, connected

domain tags

1, 2, 4, and 5 can all be recycled, and can be reused in the part below the dashed box without affecting them.

The foregoing has explained the main ideas of the recovery and reuse of connected domain labels. The following still takes Figure 9 as an example to illustrate the process of recovery and reuse of connected domain labels. Assuming that the maximum connected component label supported during the analysis of the connected component of the image to be analyzed shown in Figure 9 is set to 7, the connected component analysis method for the first 46 pixels of the image to be analyzed shown in Figure 9 can refer to the previous description, which will not be here Repeat. After analyzing the connected component of the pixel 46, that is, after assigning a new connected component label 7 to the pixel 46, the preset seven connected component labels 1 to 7 have been used, and no new connected component label can be used. Used for subsequent connected domain analysis. Therefore, at this time, the connected domain label should be recovered, and the current recovery position is recorded as the pixel 46. Please refer to FIG. 10 together. FIG. 10 is a schematic diagram of a row label information provided by an embodiment of the present invention. The relationship between the row label information and the original image in FIG. Analyzing the relationship between the images, combined with the row label information in Figure 10, it can be seen that the first 6 data in the row label information from left to right are

pixels

41, 42, 43, 44, 45, 46 after the connected domain analysis Connected domain labels 0, 0, 0, 0, 0, 7; and the next 6 data are the

pixels

37, 38, 39, 3a, 3b, 3c in the upper row of the row where the pixel 46 is after the connected domain analysis Connected domain label; 0, 0, 0, 3, 0, 6.

Further, the row label information shown in Figure 10 is sorted out once, as long as the root node label appears in the row label information, it is possible to continue to grow downward, and the root node label that does not appear in the row label information points to All connected domain labels in the connected domains can be recycled. Based on the row label information in Figure 10, it can be found that the root node labels that may continue to grow downward are 3, 7, and 6, and then the

root node labels

3, 7, and 6 are grown in the connected domain statistics. Please also refer to FIG. 11, which is a schematic diagram of connected domain statistical information provided by the second embodiment of the present invention. Specifically, the connected domain statistical information corresponding to the image to be analyzed shown in FIG. 9 is sorted before the connected domain label is recovered. Schematic. As shown in Figure 11, the first identification bit is used to indicate the type of statistical data stored in the data bit, and the second identification bit is used to record the growth identification of the connected domain tag. When the second identification bit is "0", it indicates the connected domain The label is a non-growing label. When the second identification bit is "1", it indicates that the connected domain label is a growing label. It can be seen that the connected domain label 4 points to the root node label 1 (that is, 01), the connected domain label 5 points to the root node label 2 (that is, 10), and the connected domain labels 3, 6, and 7 are growth labels. Among them, after the

root node labels

3, 7, and 6 are marked for growth in the connected domain statistical information, the connected domain statistical information is sorted, and all intermediate node labels are changed to point to the root node label. It should be noted that in this example, before and after the row label information shown in FIG. 10 is combed, the recorded connected domain statistical information remains unchanged, and they are all connected domain statistical information shown in FIG. 11. Among them, the combing process is illustrated as follows: suppose that a linked list of 10->9->8->7->6 is recorded in the connected domain statistics information, and 7->6 indicates that the connected domain label 7 points to the connected domain label 6, from the linked list It can be seen that starting from the connected component label 10, it can be determined that the connected component label 6 is the root node label, and then the connected component labels 10, 9, and 8 are all modified to the first-level intermediate node label pointing to the root node label 6, which is 10- >6,9->6,8->6,7->6.

Further, a traversal is performed on the connected component statistical information shown in FIG. 11, and the root node label whose growth mark is "0" and the first-level intermediate node labels pointing to these root node labels are determined. The above two types of connected component labels are also It is a non-growth label. By traversing the connected domain statistical information shown in Figure 11, four

non-growth tags

1, 2, 4, and 5 can be determined. At this time, the

non-growth tags

1, 2, 4, and 5 can be stored in the preset storage space. Further, based on the

non-growth labels

1, 2, 4, and 5, the subsequent connected domain analysis of the pixel to be analyzed is continued starting from the pixel 47. For the case where a new connected domain label needs to be assigned, only the preset is stored The reusable non-growth tags stored in the space can be read and used. Please also refer to FIG. 12, which is a schematic diagram of a connected domain provided by the third embodiment of the present invention. Specifically, the connected domain label recovery and reuse method described above is used to complete the connected domain analysis result of all pixels in FIG. As shown in Figure 12, the pixel point 46 and the pixel point before it, that is, the part of the grid that is not shown in bold in the figure, is the part marked by the preset 7 connected domain labels 1 to 7, that is, It is the part before the recycling and reuse of connected domain labels. Pixel 47 and the following pixels, that is, the part shown in bold squares in the figure, are the parts marked with the recycled non-growth tags, that is, the parts reused by the connected domain tags. The above two parts can be distinguished by the recorded recovery position pixels 46. Among them, the connected component labels corresponding to pixels 49, 4a, 59, 69 are non-growth label 1, and the connected component labels corresponding to pixels 52, 53, 54, 55, 56, 62, and 65 are non-growth label 2. The connected domain label corresponding to 6b and 6c is the non-growth label 4. It can be seen that the three connected domain labels 1, 2, and 4 out of the 4 non-growth labels recovered are all label reused.

In an embodiment, after the connected domain analysis of all the pixels to be analyzed in the image to be analyzed is completed, the data processing device may also determine, according to the recorded connected domain statistical information and the label map, that the image to be analyzed does not meet the preset conditions Connected domains, and filter pixels in the connected domains that do not meet the preset conditions. In one embodiment, not satisfying the preset condition means that the number of connected pixels in the connected domain is less than the preset value.

The following is an example to illustrate with the above-mentioned example. It is assumed that it is desired to filter out connected domains in the image to be analyzed that the number of connected pixels is less than 5 through connected domain analysis. In the previous process of recovering connected domain labels, after traversing the connected domain statistics shown in Figure 11, four

non-growth labels

1, 2, 4, and 5 are determined, because at this time, the connected

domain labels

1 and 4 are composed of The number of connected pixels in the connected domain is 5, then the pixels in this connected domain should be retained; at this time, the number of connected pixels in the connected domain composed of

connected domain labels

2 and 5 is 4, then this The pixels in the connected domain should be filtered out; at this time, the

non-growth tags

1, 4 can be stored in the first preset storage space, and the

non-growth tags

2, 5 can be stored in the second preset storage space. Please refer to FIG. 13 together, which is a schematic diagram of connected domain statistical information provided by the third embodiment of the present invention. Specifically, the connection recorded after the connected domain analysis of all pixels in FIG. 9 is completed using the above-mentioned connected domain label recovery and reuse method Domain statistics. As shown in FIG. 13, the data corresponding to the connected domain labels 1, 2, and 4 in the connected domain statistical information are all recorded for the part of pixels shown in bold in FIG. 12. It can be seen that connected component label 3 points to root node label 1, connected

component labels

5 and 7 both point to root node label 2, the number data corresponding to connected component label 1 is 8, the number data corresponding to connected component label 2 is 8, and the connected component label 4 The corresponding number data is 2.

Since the connected component labels in the unbold part of the square and the bold part of the square are duplicated, it is necessary to traverse the connected component statistical information shown in Figure 13 once, and use the statistical data of all intermediate node labels as the root node The statistical data of the label is replaced, otherwise it will affect the result judgment. Please refer to FIG. 14 together. FIG. 14 is a schematic diagram of a connected domain statistical information provided by a fourth embodiment of the present invention. Specifically, it is the result of the connected domain statistical information traversal shown in FIG. The connected domains in the thick part are analyzed. Since the number data corresponding to the connected component label 4 is less than 5, it is necessary to filter out the pixels with the connected component label 4 in the bold part of the grid; combine the connected component labeling result shown in Figure 12, or the recorded image to be analyzed corresponding It can be determined that the pixels with the connected domain label of 4 in the thickened part of the grid are 6b and 6c, so the pixels 6b and 6c in the image to be analyzed shown in Figure 9 can be filtered out.

For the connected component analysis of the unbolded part of the grid, since the statistical data corresponding to the reused non-growth label has been modified in the connected component statistical information shown in Figure 14, the connected component statistical information shown in Figure 14 cannot be used directly Analyzed, it is necessary to read the connected

domain tags

1, 4, 2, and 5 stored in the first preset storage unit and the second preset storage unit when the connected domain tags are recovered, and modify the data shown in Figure 14 according to the different read addresses. The corresponding statistical data in the connected domain statistical information shown. Please refer to FIG. 15 together, which is a schematic diagram of connected domain statistical information provided by the fifth embodiment of the present invention. Specifically, the connected

domain label

1, 4, 2, and 5 read out different addresses for the connected domain shown in FIG. The modified result of the corresponding statistical data in the domain statistical information. Specifically, the connected domain label read from the first preset storage unit indicates that the number of connected pixels in the corresponding connected domain is greater than or equal to 5, which cannot be filtered, and the connected domain read from the second preset storage unit The domain label indicates that the number of connected pixels in the corresponding connected domain is less than 5 and can be filtered. As shown in Figure 15, the number data in the connected domain statistical information corresponding to the connected

domain tags

1, 4 read from the first preset storage unit is directly modified to 5 (that is, 101), and the second The number data in the connected domain statistical information corresponding to the connected

domain tags

2 and 5 read from the preset storage unit is modified to 0 to facilitate subsequent judgment. Since the number data corresponding to connected domain labels 2, 5, and 6 is less than 5, it is necessary to filter out pixels with connected domain labels 2, 5, and 6 in the unbold part of the grid; combined with the connected domain labels shown in Figure 12 As a result, or the label image corresponding to the recorded image to be analyzed, it can be determined that the pixels with the connected domain labels 2, 5, and 6 in the unbold part of the grid are 16, 26, 27, 25, 3c, so the image can be changed The pixels 16, 26, 27, 25, 3c in the image to be analyzed shown in 9 are filtered out. With the above method, the pixels in the connected domain whose number of connected pixels in the image to be analyzed are less than the preset value can be filtered.

It should be noted that, due to the possibility of recovering connected domain labels once during the processing process, it is not possible to completely analyze an image to be analyzed. At this time, you can continue to perform the above connected domain label recovery and reuse process. For specific processing methods, please refer to the previous description. , I won’t repeat it here. In the connected domain analysis process, for example, in the process of filtering the pixels in the connected domain that do not meet the preset conditions, the image to be analyzed needs to be analyzed in the order from bottom to top, from right to left. Contrary to the process of determining the connected domain label of the pixel to be analyzed; this can ensure to a certain extent that the reused connected domain label does not affect the analysis result. If you also want to perform connected domain analysis in the order from top to bottom and from left to right, you need to back up the connected domain statistics at multiple stages in the processing process. The backed-up connected domain statistics can be realized from top to bottom. Analyze connected components in the order of bottom and left to right.

Refer to FIG. 16, which is a schematic structural diagram of a data processing device provided by an embodiment of the present invention. The data processing device described in the embodiment of the present invention includes a processor 1601, a communication interface 1602, and a memory 1603. Among them, the processor 1601, the communication interface 1602, and the memory 1603 may be connected through a bus or other methods. The embodiment of the present invention takes the connection through a bus as an example.

The processor 1601 may be a central processing unit (CPU), a network processor (NP), or a combination of a CPU and an NP. The processor 1601 may also be a multi-core CPU or a core used to implement communication identification binding in a multi-core NP.

The processor 1601 may be a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD) or a combination thereof. The PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a general array logic (generic array logic, GAL) or any combination thereof.

The communication interface 1602 can be used for the exchange of sending and receiving information or signaling, as well as the reception and transmission of signals. The memory 1603 may mainly include a storage program area and a storage data area. The storage program area can store an operating system and a storage program required by at least one function (such as a text storage function, a location storage function, etc.); the storage data area can store Data (such as image data, text data) created according to the use of the device, etc., and may include application storage programs, etc. In addition, the memory 1603 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices.

The memory 1603 is also used to store program instructions. The processor 1601 is configured to execute program instructions stored in the memory 1603, and when the program instructions are executed, the processor 1601 is configured to:

Acquiring an image to be analyzed, performing a connected domain analysis on the current pixel to be analyzed in the image to be analyzed, and recording a target connected domain label for the pixel to be analyzed in the label map of the image to be analyzed according to the analysis result; Update the connected component statistical information of the image to be analyzed according to the target connected component label; wherein, when the connected component statistical information is updated, if the target connected component label is the root node label, then The domain statistical information records the first type of statistical data for the target connected domain label, where the first type of statistical data is number data, and the number data is used to indicate the target connected domain label and the marked connected domain label The number of pixels pointing to the target connected domain label; if the target connected domain label is an intermediate node label, record the second type of statistical data for the target connected domain label in the connected domain statistical information, so The second type of statistical data is pointer data, and the pointer data is used to indicate the root node label pointed to by the target connected domain label.

The methods executed by the processor in the embodiments of the present invention are all described from the perspective of the processor. It can be understood that the processor in the embodiments of the present invention requires the cooperation of other hardware structures to execute the foregoing methods. The embodiments of the present invention do not make detailed descriptions and restrictions on the specific implementation process.

In an embodiment, the connected domain statistical information includes a first identification bit and a data bit, wherein the first identification bit is used to indicate the type of statistical data recorded by the data bit, and the data bit is used to indicate Is the statistical data recorded for the target connected domain label; when the first identification bit is the first identifier, the data bit is the first type of statistical data recorded for the target connected domain label, when the first When the identification bit is the second identification, the data bit is the second type of statistical data recorded by the target connected domain tag.

In an embodiment, when the processor 1601 performs a connected domain analysis on the current pixel to be analyzed in the image to be analyzed, it is specifically configured to: determine from the image to be analyzed that the pixel to be analyzed is current Adjacent adjacent pixels; performing connected domain analysis on the pixel to be analyzed according to the adjacent pixel to determine the target connected domain label of the pixel to be analyzed.

In an embodiment, the adjacent pixel points include: a first pixel point that is adjacent in the upper direction in the same column as the pixel point to be analyzed, and/or is in the same row as the pixel point to be analyzed The second pixel point adjacent in the center-left direction.

In an embodiment, when the processor 1601 performs connected domain analysis on the pixel to be analyzed according to the adjacent pixels, and determines the target connected domain label of the pixel to be analyzed, it is specifically configured to: The first pixel performs connected domain analysis on the pixel to be analyzed; if it is determined that the pixel to be analyzed is connected to the first pixel, then the connected domain label of the first pixel is used as the The target connected domain label of the pixel to be analyzed.

In an embodiment, when the processor 1601 performs connected domain analysis on the pixel to be analyzed according to the adjacent pixels, and determines the target connected domain label of the pixel to be analyzed, it is specifically configured to: The second pixel performs connected domain analysis on the pixel to be analyzed; if it is determined that the pixel to be analyzed is connected to the second pixel, then the connected domain label of the second pixel is used as the The target connected domain label of the pixel to be analyzed.

In an embodiment, when the processor 1601 performs connected domain analysis on the pixel to be analyzed according to the adjacent pixels, and determines the target connected domain label of the pixel to be analyzed, it is specifically configured to: The first pixel and the second pixel perform a connected domain analysis on the pixel to be analyzed; if it is determined that the pixel to be analyzed is connected to the first pixel and the second pixel , The connected component label of the connected component of the first pixel and the connected component of the connected component label of the second pixel are used as the target connected component label of the pixel to be analyzed.

In an embodiment, the processor 1601 is further configured to: determine the connected component label with the largest label value among the connected component labels of the first pixel and the connected component labels of the second pixel as pointing to the The intermediate node label of the target connected domain label; the connected domain label with the largest label value in the connected domain statistical information is changed.

In an embodiment, when the processor 1601 changes the connected component label with the largest label value in the connected component statistical information, it is specifically configured to: change the connected component statistical information with the largest label value. The first identification bit corresponding to the connected domain label is adjusted to the second identification; the data bit corresponding to the connected domain label with the largest label value in the connected domain statistical information is the connected domain label record with the largest label value The second type of statistical data, the recorded second type of statistical data indicates that the connected component label with the largest label value points to the target connected component label.

In one embodiment, the processor 1601 is further configured to: if it is determined that the pixel to be analyzed is not connected with the first pixel and the second pixel, connect a plurality of preset An unused connected domain label in the domain labels is used as the target connected domain label of the pixel to be analyzed.

In an embodiment, the connected domain label of the first pixel that is adjacent in the upper direction in the same column as the pixel to be analyzed is recorded in the row label information; the row label information further includes The connected domain label of the pixel in the left direction in the same row of the pixel to be analyzed, and the connected domain label of the pixel in the right direction in the same row of the first pixel.

In an embodiment, the processor 1601 is further configured to: after the connected domain analysis of the pixel to be analyzed is completed, perform the row label information on the line label information according to the determined target connected domain label of the pixel to be analyzed Update.

In an embodiment, the processor 1601 is further configured to: after a plurality of preset connected domain labels are used up, determine a growth label and a non-information from the plurality of connected domain labels according to the currently recorded row label information. Growing label; according to the non-growing label, the connected area analysis is performed on the pixels in the image to be analyzed that have not completed the connected area analysis.

In an embodiment, when the processor 1601 determines a growing label and a non-growing label from the plurality of connected domain labels according to the row label information currently recorded, it is specifically configured to: Determine the first label that exists in the line label information of the current record, and determine the first label as a growth label; determine from the plurality of connected domain labels that are not present in the line label information of the current record A second label, and determining the second label as a non-growth label.

In an embodiment, when the processor 1601 determines a growing label and a non-growing label from the plurality of connected domain labels according to the row label information currently recorded, it is specifically configured to: Determine the first label that exists in the line label information of the current record, and determine the second label that does not exist in the line label information of the current record; according to the connected domain statistical information, from the plurality of connected domain labels Is determined as the root node label of the first label or the third label of the intermediate node label; the labels other than the third label in the second label are determined as non-growth labels, and the first label The label and the third label are determined to be growth labels.

In an embodiment, the connected domain statistical information includes a second identification bit, the second identification bit is used to record the growth identification of the connected domain label, and when the second identification bit is the third identification, it indicates the connected domain label It is a non-growth tag, and when the second identification bit is a fourth tag, it indicates that the connected domain tag is a growth tag.

In an embodiment, the processor 1601 is further configured to: determine, according to the connected domain statistical information and the label map, a connected domain that does not meet a preset condition in the image to be analyzed; Set the pixels in the connected domain of the condition to filter.

In one embodiment, not satisfying the preset condition means that the number of connected pixels in the connected domain is less than the preset value.

In an embodiment, the connected domain statistical information is stored in an on-chip random access memory RAM, and/or the row label information is stored in an on-chip register, and/or, the label map is stored in a chip External double-rate synchronous dynamic random access memory DDR.

In specific implementation, the processor 1601, the communication interface 1602, and the memory 1603 described in the embodiment of the present invention can execute the implementation described in the method for processing a connected domain provided in the embodiment of the present invention, and details are not described herein again.

An embodiment of the present invention also provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the connected domain processing method described in the foregoing method embodiment is implemented.

The embodiment of the present invention also provides a computer program product containing instructions, which when run on a computer, causes the computer to execute the connected domain processing method described in the foregoing method embodiment.

It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described sequence of actions. Because according to the present invention, certain steps can be performed in other order or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the involved actions and modules are not necessarily required by the present invention.

A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above-mentioned embodiments can be completed by a program instructing relevant hardware. The program can be stored in a computer-readable storage medium, and the storage medium can include: Flash disk, read-only memory (Read-Only Memory, ROM), random access device (Random Access Memory, RAM), magnetic disk or optical disk, etc.

The above describes in detail a connected domain processing method and data processing equipment provided by the embodiments of the present invention. Specific examples are used in this article to explain the principles and implementation of the present invention. The description of the above embodiments is only for help Understand the method of the present invention and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation and the scope of application. It should be understood as a limitation of the present invention.

Claims

A connected domain processing method, characterized in that the method includes:

Acquiring an image to be analyzed, performing a connected domain analysis on the current pixel to be analyzed in the image to be analyzed, and recording a target connected domain label for the pixel to be analyzed in the label map of the image to be analyzed according to the analysis result;

Updating the connected domain statistical information of the image to be analyzed according to the target connected domain label;

Wherein, when the connected component statistical information is updated, if the target connected component label is a root node label, then the first type of statistical data is recorded for the target connected component label in the connected component statistical information, so The first type of statistical data is number data, and the number data is used to indicate the number of pixels marked as the target connected domain label and the marked connected domain label pointing to the target connected domain label; if the target is connected If the domain label is an intermediate node label, then the second type of statistical data is recorded for the target connected domain label in the connected domain statistical information, the second type of statistical data is pointer data, and the pointer data is used to indicate the The label of the root node pointed to by the target connected domain label.
The method according to claim 1, wherein the connected domain statistical information includes a first identification bit and a data bit, wherein the first identification bit is used to indicate the type of statistical data recorded by the data bit, The data bit is used to represent the statistical data recorded for the target connected domain label; when the first identification bit is the first identifier, the data bit is the first type of statistics recorded for the target connected domain label Data. When the first identification bit is the second identification, the data bit records the second type of statistical data for the target connected domain tag.
The method according to claim 2, wherein the performing connected domain analysis on the current pixel to be analyzed in the image to be analyzed comprises:

Determining adjacent pixels adjacent to the current pixel to be analyzed from the image to be analyzed;

A connected domain analysis is performed on the pixel to be analyzed according to the adjacent pixels, and a target connected domain label of the pixel to be analyzed is determined.
The method according to claim 3, wherein the adjacent pixel points comprise: the first pixel point adjacent to the upper direction in the same column as the pixel point to be analyzed, and/or, and the The pixel to be analyzed is the second pixel adjacent to the left in the same row.
The method according to claim 4, wherein the performing connected domain analysis on the pixel to be analyzed according to the neighboring pixels to determine the target connected domain label of the pixel to be analyzed comprises:

Performing connected domain analysis on the pixel to be analyzed according to the first pixel;

If it is determined that the pixel to be analyzed is connected to the first pixel, the connected region label of the first pixel is used as the target connected region label of the pixel to be analyzed.
The method according to claim 4, wherein the performing connected domain analysis on the pixel to be analyzed according to the neighboring pixels to determine the target connected domain label of the pixel to be analyzed comprises:

Performing connected domain analysis on the pixel to be analyzed according to the second pixel;

If it is determined that the pixel to be analyzed is connected to the second pixel, the connected region label of the second pixel is used as the target connected region label of the pixel to be analyzed.
The method according to claim 4, wherein the performing connected domain analysis on the pixel to be analyzed according to the neighboring pixels to determine the target connected domain label of the pixel to be analyzed comprises:

Performing connected domain analysis on the pixel to be analyzed according to the first pixel and the second pixel;

If it is determined that the pixel to be analyzed is connected to the first pixel and the second pixel, then the connected domain label of the first pixel and the connected domain label of the second pixel are The connected component label with the smallest label value is used as the target connected component label of the pixel to be analyzed.
The method according to claim 7, wherein the method further comprises:

Determining the connected component label with the largest label value among the connected component label of the first pixel and the connected component label of the second pixel as an intermediate node label pointing to the target connected component label;

Change the connected domain label with the largest label value in the connected domain statistical information.
The method according to claim 8, wherein the changing the connected domain label with the largest label value in the connected domain statistical information comprises:

Adjusting the first identification bit corresponding to the connected domain label with the largest label value in the connected domain statistical information to the second identification;

In the connected domain statistical information, the data bit corresponding to the connected domain label with the largest label value is recorded for the connected domain label with the largest label value, and the recorded second type statistical data indicates the label The connected component label with the largest value points to the target connected component label.
The method according to any one of claims 7 to 9, wherein the method further comprises:

If it is determined that the pixel to be analyzed is not connected to the first pixel and the second pixel, then an unused connected domain label among a plurality of preset connected domain labels is used as the to-be-analyzed pixel. Analyze the target connected domain label of the pixel.
The method according to any one of claims 4 to 10, wherein the connected domain label of the first pixel that is adjacent in the upper direction in the same column as the pixel to be analyzed is recorded in the row label information The line label information also includes the connected domain label of the pixel located in the left direction in the same row of the pixel to be analyzed, and the connected domain label of the pixel located in the right direction in the same row of the first pixel Domain label.
The method of claim 11, wherein the method further comprises:

After the connected domain analysis of the pixel to be analyzed is completed, the row label information is updated according to the determined target connected domain label of the pixel to be analyzed.
The method of claim 12, wherein the method further comprises:

After the multiple preset connected domain labels are used up, determine the growing label and the non-growing label from the multiple connected domain labels according to the row label information currently recorded;

Perform connected domain analysis on pixels in the image to be analyzed that have not completed connected domain analysis according to the non-growth tag.
The method according to claim 13, wherein the determining a growing label and a non- growing label from the plurality of connected domain labels according to the row label information of the current record comprises:

Determine the first label existing in the row label information of the current record from the plurality of connected domain labels, and determine the first label as a growth label;

A second label that does not exist in the row label information of the current record is determined from the plurality of connected domain labels, and the second label is determined as a non-growing label.
The method according to claim 13, wherein the determining a growing label and a non- growing label from the plurality of connected domain labels according to the row label information of the current record comprises:

Determine the first label existing in the row label information of the current record from the plurality of connected domain labels, and determine the second label that does not exist in the row label information of the current record;

Determining, from the plurality of connected domain labels, a third label that is the root node label of the first label or the intermediate node label according to the connected domain statistical information;

The labels other than the third label among the second labels are determined as non-growth labels, and the first label and the third label are determined as growth labels.
The method according to any one of claims 13 to 15, wherein the connected domain statistical information includes a second identification bit, and the second identification bit is used to record the growth identification of the connected domain label. When the second identification bit is the third identification, it indicates that the connected domain label is a non-growing label, and when the second identification bit is the fourth identification, it indicates that the connected domain label is a growing label.
The method according to any one of claims 1 to 16, wherein the method further comprises:

Determine, according to the connected domain statistical information and the label map, connected domains that do not meet a preset condition in the image to be analyzed;

Filtering the pixels in the connected domain that do not meet the preset condition.
The method according to claim 17, wherein the failure to satisfy the preset condition means that the number of connected pixels in the connected domain is less than the preset value.
The method according to claim 11 or 12, wherein the connected domain statistical information is stored in an on-chip random access memory (RAM), and/or the row label information is stored in an on-chip register, and/ Or, the label image is stored in an off-chip double-rate synchronous dynamic random access memory DDR.
A data processing device, characterized by comprising: a memory and a processor,

The memory is used to store program instructions;

The processor is configured to execute program instructions stored in the memory, and when the program instructions are executed, the processor is configured to:

Acquiring an image to be analyzed, performing a connected domain analysis on the current pixel to be analyzed in the image to be analyzed, and recording a target connected domain label for the pixel to be analyzed in the label map of the image to be analyzed according to the analysis result;

Updating the connected domain statistical information of the image to be analyzed according to the target connected domain label;

Wherein, when the connected component statistical information is updated, if the target connected component label is a root node label, then the first type of statistical data is recorded for the target connected component label in the connected component statistical information, so The first type of statistical data is number data, and the number data is used to indicate the number of pixels marked as the target connected domain label and the marked connected domain label pointing to the target connected domain label; if the target is connected If the domain label is an intermediate node label, then the second type of statistical data is recorded for the target connected domain label in the connected domain statistical information, the second type of statistical data is pointer data, and the pointer data is used to indicate the The label of the root node pointed to by the target connected domain label.
The data processing device according to claim 20, wherein the connected domain statistical information includes a first identification bit and a data bit, wherein the first identification bit is used to indicate the statistical data recorded by the data bit Type, the data bit is used to indicate the statistical data recorded for the target connected domain label; when the first identification bit is the first identifier, the data bit is the first Type statistical data, when the first identification bit is the second identification, the data bit records the second type of statistical data for the target connected domain tag.
The data processing device according to claim 21, wherein when the processor performs connected domain analysis on the current pixel to be analyzed in the image to be analyzed, it is specifically configured to:

Determining adjacent pixels adjacent to the current pixel to be analyzed from the image to be analyzed;

A connected domain analysis is performed on the pixel to be analyzed according to the adjacent pixels, and a target connected domain label of the pixel to be analyzed is determined.
The data processing device according to claim 22, wherein the adjacent pixel points comprise: a first pixel point adjacent to the upper direction in the same column as the pixel to be analyzed, and/or, and The pixel to be analyzed is a second pixel adjacent to the left in the same row.
The data processing device according to claim 23, wherein the processor performs a connected domain analysis on the pixel to be analyzed according to the adjacent pixels to determine the target connected domain of the pixel to be analyzed When labeling, specifically used for:

Performing connected domain analysis on the pixel to be analyzed according to the first pixel;

If it is determined that the pixel to be analyzed is connected to the first pixel, the connected region label of the first pixel is used as the target connected region label of the pixel to be analyzed.
The data processing device according to claim 23, wherein the processor performs a connected domain analysis on the pixel to be analyzed according to the adjacent pixels to determine the target connected domain of the pixel to be analyzed When labeling, specifically used for:

Performing connected domain analysis on the pixel to be analyzed according to the second pixel;

If it is determined that the pixel to be analyzed is connected to the second pixel, the connected region label of the second pixel is used as the target connected region label of the pixel to be analyzed.
The data processing device according to claim 23, wherein the processor performs a connected domain analysis on the pixel to be analyzed according to the adjacent pixels to determine the target connected domain of the pixel to be analyzed When labeling, specifically used for:

Performing connected domain analysis on the pixel to be analyzed according to the first pixel and the second pixel;

If it is determined that the pixel to be analyzed is connected to the first pixel and the second pixel, then the connected domain label of the first pixel and the connected domain label of the second pixel are The connected component label with the smallest label value is used as the target connected component label of the pixel to be analyzed.
The data processing device according to claim 26, wherein the processor is further configured to:

Determining the connected component label with the largest label value among the connected component label of the first pixel and the connected component label of the second pixel as an intermediate node label pointing to the target connected component label;

Change the connected domain label with the largest label value in the connected domain statistical information.
The data processing device according to claim 27, wherein when the processor changes the connected domain label with the largest label value in the connected domain statistical information, it is specifically configured to:

Adjusting the first identification bit corresponding to the connected domain label with the largest label value in the connected domain statistical information to the second identification;

In the connected domain statistical information, the data bit corresponding to the connected domain label with the largest label value is recorded for the connected domain label with the largest label value, and the recorded second type statistical data indicates the label The connected component label with the largest value points to the target connected component label.
The data processing device according to any one of claims 26 to 28, wherein the processor is further configured to:

If it is determined that the pixel to be analyzed is not connected to the first pixel and the second pixel, then an unused connected domain label among a plurality of preset connected domain labels is used as the to-be-analyzed pixel. Analyze the target connected domain label of the pixel.
The data processing device according to any one of claims 23 to 29, wherein the connected domain label of the first pixel that is adjacent in the upper direction in the same column as the pixel to be analyzed is recorded in the row In the label information; the row label information also includes a connected domain label of a pixel located in the left direction in the same row of the pixel to be analyzed, and a pixel located in the right direction in the same row of the first pixel The connected domain label.
The data processing device according to claim 30, wherein the processor is further configured to:

After the connected domain analysis of the pixel to be analyzed is completed, the row label information is updated according to the determined target connected domain label of the pixel to be analyzed.
The data processing device according to claim 31, wherein the processor is further configured to:

After the multiple preset connected domain labels are used up, determine the growing label and the non-growing label from the multiple connected domain labels according to the row label information currently recorded;

The connected area analysis is performed on the pixels in the image to be analyzed that have not completed the connected area analysis according to the non-growth label.
The data processing device according to claim 32, wherein when the processor determines a growing label and a non-growing label from the plurality of connected domain labels according to the row label information currently recorded, it is specifically configured to:

Determine the first label existing in the row label information of the current record from the plurality of connected domain labels, and determine the first label as a growth label;

A second label that does not exist in the row label information of the current record is determined from the plurality of connected domain labels, and the second label is determined as a non-growing label.
The data processing device according to claim 32, wherein when the processor determines a growing label and a non-growing label from the plurality of connected domain labels according to the row label information currently recorded, it is specifically configured to:

Determine the first label existing in the row label information of the current record from the plurality of connected domain labels, and determine the second label that does not exist in the row label information of the current record;

Determining, from the plurality of connected domain labels, a third label that is the root node label of the first label or the intermediate node label according to the connected domain statistical information;

The labels other than the third label among the second labels are determined as non-growth labels, and the first label and the third label are determined as growth labels.
The data processing device according to any one of claims 32 to 34, wherein the connected domain statistical information includes a second identification bit, and the second identification bit is used to record the growth identifier of the connected domain label, when When the second identification bit is a third identification, it indicates that the connected domain label is a non-growth tag, and when the second identification bit is a fourth identification, it indicates that the connected domain label is a growing tag.
The data processing device according to any one of claims 20 to 35, wherein the processor is further configured to:

Determine, according to the connected domain statistical information and the label map, connected domains that do not meet a preset condition in the image to be analyzed;

Filtering the pixels in the connected domain that do not meet the preset condition.
The data processing device according to claim 36, wherein the failure to satisfy the preset condition means that the number of connected pixels in the connected domain is less than a preset value.
The data processing device according to claim 30 or 31, wherein the connected domain statistical information is stored in an on-chip random access memory (RAM), and/or the row label information is stored in an on-chip register, And/or, the label map is stored in an off-chip double-rate synchronous dynamic random access memory DDR.
A computer-readable storage medium in which a computer program is stored, characterized in that: when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 19 are implemented .