CN110264488A - A kind of bianry image edge extraction device - Google Patents
A kind of bianry image edge extraction device Download PDFInfo
- Publication number
- CN110264488A CN110264488A CN201910535278.1A CN201910535278A CN110264488A CN 110264488 A CN110264488 A CN 110264488A CN 201910535278 A CN201910535278 A CN 201910535278A CN 110264488 A CN110264488 A CN 110264488A
- Authority
- CN
- China
- Prior art keywords
- unit
- pixel
- edge
- data
- row
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000605 extraction Methods 0.000 title claims abstract description 32
- 238000012545 processing Methods 0.000 claims abstract description 41
- 239000011159 matrix material Substances 0.000 claims abstract description 35
- 239000000284 extract Substances 0.000 claims description 9
- 238000013075 data extraction Methods 0.000 claims description 2
- 238000004422 calculation algorithm Methods 0.000 description 7
- 238000000034 method Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000003909 pattern recognition Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000001154 acute effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000007634 remodeling Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/60—Memory management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
This application discloses a kind of bianry image edge extraction device, which includes: acquiring unit, storage unit, processing unit and image output unit；Acquiring unit is used to obtain the bianry image of images to be recognized, and according to the rank addresses of pixel in bianry image, the point-by-point pixel number evidence for obtaining pixel；Storage unit is used to according to three dimensional neighborhood matrix form, store pixel number evidence, and export to processing unit according to rank addresses；Processing unit is used to, to receive, three dimensional neighborhood matrix form, pixel number according to Edge Gradient Feature operation is carried out, the result of Edge Gradient Feature operation is denoted as edge feature value, and edge feature value is input to image output unit according to rank addresses；Image output unit is used to generate and export the edge feature image of images to be recognized.By the technical solution in the application, the complexity of edge extracting operation is reduced, data space is saved.
Description
Technical field
This application involves the technical fields of image processing apparatus, fill in particular to a kind of bianry image edge extracting
It sets.
Background technique
As Information Technology Development is advanced by leaps and bounds, image processing techniques also develops rapidly therewith, extensively at present
Every aspect applied to daily life.In Digital Image Processing, it is image that edge feature, which is an important feature of image,
The important component of processing, pattern-recognition and computer vision, is just being widely used in the extraction operation of picture edge characteristic
The fields such as industrial detection, image segmentation, motion detection, recognition of face and target following, and to the result of Edge extraction
Also the effect of further image procossing and pattern-recognition will be directly affected.
The edge of image is usually the place that acute variation occurs for color of image, and these variations are often by the shape of object
Caused by light is reflected on the surface of shape structure, external environment illumination and object.The edge of image can directly reflect the wheel of object
Wide and topological structure, Edge extraction technology are the important bases of image processing techniques, mode identification technology, computation vision technology
One of plinth.With the development in nowadays fields such as artificial intelligence and big data cloud computing, speed and figure of the people to machine recognition
The rate request of picture processing is higher and higher, that is to say, the bright speed to Edge extraction has also begun with and cannot be neglected pass
Note.
It in the prior art, is all to carry out directly carrying out edge extraction operation after gray processing processing to color image mostly,
Although the data after gray processing are 8bit data, its numerical value or bigger causes when carrying out edge extracting still
Maintain higher computation complexity.
Summary of the invention
The purpose of the application is: a kind of bianry image edge extraction device is realized, to reduce Edge Gradient Feature operation
Complexity, reduce memory space, and then realize the extraction to picture edge characteristic quickly.
The technical solution of the application is: providing a kind of bianry image edge extraction device, which includes: to obtain list
Member, storage unit, processing unit and image output unit；Acquiring unit is used to obtain the bianry image of images to be recognized, and
According to the rank addresses of pixel in bianry image, the point-by-point pixel number evidence for obtaining pixel；Storage unit is used for according to row
Column address stores pixel number evidence according to three dimensional neighborhood matrix form, and exports to processing unit；Processing unit is used for basis
Rank addresses, to receive, three dimensional neighborhood matrix form, pixel number according to Edge Gradient Feature operation is carried out, by edge
The result of feature extraction operation is denoted as edge feature value, and edge feature value is input to image output unit；Image output is single
Member is for generating and exporting the edge feature image of images to be recognized.
In any of the above-described technical solution, further, acquiring unit includes: line count device and judging unit；Ranks
Counter is set to the input terminal of acquiring unit, and line count device is based on carrying out the pixel in the bianry image got
Number, and generate rank addresses, wherein rank addresses include row address column address and column address row address；Judging unit is for working as
When determining that the row address column address of pixel is equal to the first preset threshold or column address row address, is equal to one or two preset threshold,
The pixel number evidence of pixel is delivered to image output unit.
In any of the above-described technical solution, further, storage unit include the first memory module, the second memory module with
And level-one register, wherein the first memory module and the second memory module are first-in first-out type memory；The data of storage unit
Input terminal is connected to the input terminal of the first memory module and the input terminal of level-one register；The output end of level-one register connects
It is connected to the third output end of storage unit；The output end of first memory module be connected to storage unit second output terminal and
The input terminal of second memory module；The output end of second memory module is connected to the first output end of storage unit.
In any of the above-described technical solution, further, processing unit includes: data extracting unit and arithmetic element；Number
It is used to mention when determining that row address column address and column address row address are all larger than or are equal to two or three preset threshold according to extraction unit
It takes and sends the pixel number of three dimensional neighborhood matrix form according to arithmetic element；Arithmetic element is used for according to pixel number according to progress
Edge Gradient Feature operation generates edge feature value.
In any of the above-described technical solution, further, arithmetic element further include: first adder, second adder, the
Three adders, the first shift unit, the second shift unit and comparing unit；First adder and second adder are set parallel
It sets, the data output end of first adder and second adder passes through the first shift unit and the second shift unit, connection respectively
Comparing unit is connected in the data output end of the data input pin of third adder, third adder；Comparing unit is used for root
According to the size between the operation result and edge preset threshold value of third adder, Edge Gradient Feature is carried out, exports edge feature
Value.
In any of the above-described technical solution, further, extracts and send three dimensional neighborhood matrix form pixel number according to extremely
Arithmetic element specifically includes:
Data decimation frame is generated, data decimation frame is 3 × 3 matrix forms；
The pixel number evidence that will be exported according to three dimensional neighborhood matrix form, is delivered to data decimation frame；
The successively pixel number of the first row first row of extraction data decimation frame, the first row secondary series, the second row first row
According to being delivered to first adder；
Successively extract the second row third column, the third line secondary series, the tertial pixel number of the third line of data decimation frame
According to being delivered to second adder.
The beneficial effect of the application is:
Before Edge Gradient Feature operation, obtain images to be recognized is bianry image, that is to say, that the number of pixel
There was only 1bit according to digit, reduces the calculation amount in Edge Gradient Feature calculating process, greatly simplified calculating process.
During the storage and processing to image, in conjunction with the storage unit for being provided with level-one register, using caching two
The mode of row pixel data constitutes three dimensional neighborhood matrix, largely realizes using less resource and reach higher storage
Efficiency.Answering for operation is reduced to pixel number according to Edge Gradient Feature operation is carried out using 3 × 3 three dimensional neighborhood matrix form
Miscellaneous degree, can more efficient quickly finish the Edge Gradient Feature to image.
In the application, in the boundary pixel point processing mode to image, do not use to pixel carry out symmetric extension and
The processing of periodic extension, but same position is directly indicated with the pixel value of the initial pictures of same position (pixel number evidence)
Result pixel value (edge feature value), eliminate the operation to these pixels, enormously simplify the storage to whole frame picture
And calculating operation.
Detailed description of the invention
The advantages of above-mentioned and/or additional aspect of the application, will become bright in combining description of the following accompanying drawings to embodiment
It shows and is readily appreciated that, in which:
Fig. 1 is the schematic block diagram according to the bianry image edge extraction device of one embodiment of the application；
Fig. 2 is the state machine diagram according to one embodiment of the application；
Fig. 3 is the schematic diagram that data store in the storage unit according to one embodiment of the application；
Fig. 4 is the schematic diagram according to the arithmetic element of one embodiment of the application；
Fig. 5 is the schematic diagram according to the images to be recognized of one embodiment of the application；
Fig. 6 is the schematic diagram according to the edge feature image of one embodiment of the application.
Specific embodiment
It is with reference to the accompanying drawing and specific real in order to be more clearly understood that the above objects, features, and advantages of the application
Mode is applied the application is further described in detail.It should be noted that in the absence of conflict, the implementation of the application
Feature in example and embodiment can be combined with each other.
In the following description, many details are elaborated in order to fully understand the application, still, the application may be used also
To be implemented using other than the one described here other modes, therefore, the protection scope of the application is not by described below
Specific embodiment limitation.
In the present embodiment, set the size of images to be recognized as 640*480, i.e., it is point-by-point to the image in the memory unit
When being stored, pixel column address row_count=1,2 ..., 640, pixel row address col_count=1,2 ...,
480。
As shown in Figure 1, present embodiments providing a kind of bianry image edge extraction device, comprising: acquiring unit 10, storage
Unit 20, processing unit 30 and image output unit 40；
Acquiring unit 10 is used to obtain the bianry image of images to be recognized, and according to the ranks of pixel in bianry image
Location, the point-by-point pixel number evidence for obtaining pixel；
Specifically, by prior art, the pixel value of pixel in images to be recognized can be converted into 8bit, and
Images to be recognized after conversion is denoted as gray level image, in order to reduce data volume, reduce computational complexity, acquiring unit 10 is raw
The size of pixel point marquee, marquee can calculate demand setting, such as 21*21 according to practical.
Using ergodic algorithm, marquee is moved line by line, and calculates the pixel mean value of pixel in marquee, is denoted as pixel threshold
Value, the pixel value of the corresponding pixel of marquee central point is compared with pixel threshold, when judgement pixel value is greater than pixel
When threshold value, the pixel value of the pixel is set as 1, otherwise, the pixel value of the pixel is set as 0, so, it can be by picture
The gray level image that element value is 8bit is converted to the bianry image that pixel value is 1bit, realizes and obtains the two of images to be recognized
It is worth image.
Further, acquiring unit 10 includes: line count device 11 and judging unit 12；
Line count device 11 is set to the input terminal of acquiring unit 10, and line count device 11 is used for the binary map got
Pixel as in is counted, and generates rank addresses, wherein rank addresses include column address and row address；
Judging unit 12 is used for when the column address for determining pixel is pre- equal to second equal to the first preset threshold or row address
If when threshold value, the pixel number evidence of pixel is delivered to image output unit 40.
Storage unit 20 is used to according to three dimensional neighborhood matrix form, store pixel number evidence, and export according to rank addresses
To processing unit 30；
Specifically, as shown in Fig. 2, controlling the operation processing of entire module by a finite state machine, including to picture
Reading, storage and the calculation process of vegetarian refreshments data.It includes following state that the finite state machine, which has altogether:
Low level reset signal rst_en, the signal are determined by the external input of edge extraction device 100；
Data valid signal data_en is read in, which is determined by the acquiring unit 10 of edge extraction device 100；
Into calculating section initial signal cal_en, which is determined by line count device 11 according to counting, i.e., when column ground
Location 2 < row_count < 641 and when 2 < col_count of row address < 481, the data of storage unit 20 enter processing unit 30, i.e.,
Since the 3rd column pixel number of the 3rd row according to, until the last one pixel number evidence, enter in processing unit 30, in order to
Calculate the 2nd row the 2nd column, to the 639th row the 479th arrange pixel edge feature value；
Pixel handles initial signal s_en at image boundary, which is determined by line count device 11 according to counting, i.e.,
Column address row_count=1 or column address row_count=640 or row address col_count=1 or row address
When col_count=480, the pixel number evidence of pixel is delivered to image output unit 40 by acquiring unit 10, i.e., by the 1st
The pixel number that row, the 640th row, the 1st column and the 480th arrange is delivered to output unit 40 according to as edge feature value.It needs to illustrate
, as column address row_count=1, row address col_count=1, do not execute output operation, only enter data into
Storage unit 20, wherein set the first preset threshold as 1 and 640, the second preset threshold is 1 and 480；
The data enable signal s_last_en of footline and terminal column is handled, the signal is by line count device 11 according to counting
When decision, i.e. column address row_count=640 and row address col_count=480, completion mentions the edge of images to be recognized
It takes.
Further, storage unit 20 includes the first memory module 21, the first memory module 22 and level-one register 23,
Wherein, the first memory module 21 and the first memory module 22 are first-in first-out type memory；The data input pin of storage unit 20
It is connected to the input terminal of the first memory module 21 and the input terminal of level-one register 23；The output end of level-one register 23 connects
It is connected to the third output end of storage unit 20；It is second defeated to be connected to storage unit 20 for the output end of first memory module 21
The input terminal of outlet and the first memory module 22；The output end of first memory module 22 is connected to the first output of storage unit 20
End.
Specifically, as shown in figure 3, data input pin of the setting data register data_in as storage unit 30, uses
Two identical first-in first-out type memory fifo as the first memory module fifo1 and the second memory module fifo0,
In, the depth of memory fifo is 640, width is 1, due to the depth and images to be recognized of each memory fifo
The picture traverse of bianry image is equal, and therefore, each memory fifo can store the pixel number evidence of a line bianry image.
In the present embodiment, the 32nd row 6 column datas adjacent into the 34th row are chosen, edge extraction process is said
It is bright, as shown in table 1.
Table 1
| 32nd row | … | 1 | 1 | 1 | 0 | 0 | 0 | … |
| 33rd row | … | 0 | 1 | 1 | 0 | 0 | 0 | … |
| 34th row | … | 1 | 1 | 1 | 0 | 0 | 0 | … |
Obtain the pixel number evidence of pixel point by point by acquiring unit 10, i.e., each period reads a pixel number
It is according to, the value successively obtained ..., 1,1,1,0,0,0 ..., 0,1,1,0,0,0 ..., 1,1,1,0,0,0 ....In storage unit
When 20 progress data storage, the pixel number evidence of the 32nd row first is successively read by data register data_in, and will be read simultaneously
Each the pixel number evidence got, the first memory module fifo1 is successively sent to according to reading order.When reading the 33rd row
The pixel number of 1st column according to when, all 640 pixel number evidences being stored in the first memory module fifo1 in the 32nd row, then
The pixel number evidence of the 32nd row is successively stored to the second memory module fifo0 by the first memory module fifo1.When reading
The pixel number of 34 row first rows according to when, the pixel number evidence of the 33rd row, the second storage are stored in the first memory module fifo1
The pixel number evidence of the 32nd row is stored in module fifo0.Therefore, pass through the data output end in data register data_in
Level-one register data_in_r is set, to pixel number according to the delay for carrying out a cycle, to realize the second memory module
Fifo0, the first memory module fifo1 and level-one register data_in_r export the same column data in adjacent three rows data, i.e.,
According to three dimensional neighborhood matrix form, one 3 × 3 picture element matrix is constituted, such as:
Herein it should be noted that the acquiring unit 10 of bianry image edge extraction device 100 receives images to be recognized
Bianry image after, start it is point-by-point obtain pixel number evidence, generate and read in data valid signal data_en, it is every to obtain one
Pixel number according to when, line count device 11 starts counting, firstly, row address col_count=1, column address row_count is by 1
640 are gradually increased to, then, row address col_count=2, column address row_count is gradually increased by 1 again, until reading
The last one pixel number evidence, at this time column address row_count=640, row address col_count=480, that is to say, that every
The rank addresses of one pixel number evidence can be write as the form of coordinate, the coordinate are as follows: (col_count, row_count).
The first preset threshold is set as 1 and 640, the second preset threshold is 1 and 480, that is to say, that when judging unit 12 is sentenced
When to determine pixel be at least one of following four situation, which is labeled as boundary point, corresponding pixel number evidence
As edge feature value, it is compared with edge preset threshold value.Acquiring unit 10 generates pixel processing starting at image boundary
Signal s_en is fed directly to image output unit 40 by the pixel via storage unit 20.These four situations are as follows: column ground
Location row_count=1, column address row_count=640, row address col_count=1, row address col_count=480.
It should be noted that when, as column address row_count=640 and row address col_count=480, acquiring unit
10 generate the data enable signal s_last_en for having handled footline and terminal column.
When pixel is not boundary point, acquiring unit 10 is generated into calculating section initial signal cal_en, pixel
Data enter processing unit 30 and are handled, then by treated, data are delivered to image output unit 40 by processing unit 30.
Processing unit 30 is used for according to every trade address, to receiving, three dimensional neighborhood matrix form, pixel number according into
The result of Edge Gradient Feature operation is denoted as edge feature value, and edge feature value is inputted by row Edge Gradient Feature operation
To image output unit 40；Image output unit 40 is used to generate and export the edge feature image of images to be recognized.
In this example, using the method based on Sobel operator, to the pixel number of entire bianry image according to transporting
It calculates, wherein two kinds of situations are divided into the processing of pixel, first is that the pixel at image boundary, i.e. boundary point；Second is that removing figure
As the pixel of boundary.The first situation includes the first row, first row, last line, last pixel arranged, to them
Operation be convolution algorithm without Sobel operator, and directly take the pixel number evidence of its pixel, be worth as a result, join
With Edge Gradient Feature；Second situation is the Neighborhood matrix to pixel, i.e. three dimensional neighborhood matrix, is carried out and both direction
The convolution operation of Sobel operator matrix exports the ranks of result at this point, output result is the central value of three dimensional neighborhood matrix
Address is (col_count-1, row_count-1).Therefore, in conjunction with boundary point and output as a result, in available bianry image
The corresponding edge feature value of all pixels point.
By the pixel number evidence that input data in this present embodiment is in bianry image, numerical value is 1 or 0, therefore,
In order to save calculation resources, the convolution operator of both direction can be merged into operation, calculation formula is as follows:
And then according to the matrix after merging, the data in three dimensional neighborhood matrix are extracted, using adder calculator and
The mode that shift unit combines improves arithmetic speed, reduces computational complexity.
Further, processing unit 30 includes: data extracting unit 31 and arithmetic element；Data extracting unit 31 is for working as
When determining that column address is greater than or equal to third predetermined threshold value and row address is greater than third predetermined threshold value, extracts and send three-dimensional neighbour
The pixel number of domain matrix form is according to arithmetic element；
Specifically, third predetermined threshold value is set, exactly generates three-dimensional neighbour enough to determine to be stored in storage unit 20
The pixel number evidence of domain matrix, that is to say, that only when acquiring unit 10 get the tertial pixel number of the third line according to when,
The first pixel number evidence into the third line in first three columns can be utilized in storage unit 20, generate first three dimensional neighborhood square
Battle array, meanwhile, in order to expand the data processing range of bianry image edge extraction device 100, set the value of third predetermined threshold value
It is 3.
Further, arithmetic element further include: first adder 32, second adder 33, third adder 34, first
Shift unit 35, the second shift unit 36 and comparing unit 37；First adder 32 and second adder 33 are arranged parallel, the
The data output end of one adder 32 and second adder 33, respectively by the first shift unit 35 and the second shift unit 36,
It is connected to the data input pin of third adder 34, the data output end of third adder 34 is connected to comparing unit 37；Compare
Unit 37 is used to carry out edge feature according to the size between the operation result and edge preset threshold value of third adder 34 and mention
It takes, exports edge feature value, wherein edge preset threshold value is 0.
In a kind of implementation of the present embodiment, extracts and send three dimensional neighborhood matrix form pixel number evidence to operation
Unit specifically includes:
Data decimation frame is generated, data decimation frame is 3 × 3 matrix forms；
The pixel number evidence that will be exported according to three dimensional neighborhood matrix form, is delivered to data decimation frame；
The successively pixel number of the first row first row of extraction data decimation frame, the first row secondary series, the second row first row
According to being delivered to first adder 32；
Successively extract the second row third column, the third line secondary series, the tertial pixel number of the third line of data decimation frame
According to being delivered to second adder 33.
Arithmetic element is used to generate edge feature value according to Edge Gradient Feature operation is carried out according to pixel number.
Specifically, as shown in figure 4, still with the pixel data instance of the 32nd row to the 34th row, when acquiring unit 10 obtains
To the 34th row the 4th arrange pixel number according to when, the data decimation frame of generation is as follows:
Wherein, corresponding the 1st column pixel number evidence of 32nd row of data_00, corresponding the 2nd column pixel number of 32nd row of data_01
According to, corresponding the 3rd column pixel number evidence of 32nd row of data_02, corresponding the 1st column pixel number evidence of 33rd row of data_10, data_11
Corresponding the 2nd column pixel number evidence of 33rd row, corresponding the 3rd column pixel number evidence of 33rd row of data_12, corresponding 34th row of data_20
1st column pixel number evidence, corresponding the 2nd column pixel number evidence of 34th row of data_21, corresponding the 3rd column pixel of 34th row of data_22
Data.
Calculation formula after merging further according to Sobel convolution operator, according to each pixel number in data decimation frame according to correspondence
Position, carry out data extraction, the data in data_00, data_01 and data_10 are delivered to first adder 32, will
Data in data_12, data_21 and data_22 are delivered to second adder 33, carry out add operation respectively, then to addition
Operation result carries out shifting function by the first shift unit 35 and the second shift unit 36, wherein third adder 34 executes
Subtraction.
It should be noted that since the image of input is bianry image, so using 0 value as judging whether it is with boundary
The edge preset threshold value of feature.When the operation result of both direction is 0, indicate that the point does not all have side in two directions
Edge feature indicates that the point is background dot so being assigned a value of 1；It is not 0 when the operation result of both direction has one, that is, indicates
The point has boundary characteristic, i.e. tax edge feature value is 0.
The result for carrying out Sobel convolution algorithm to three dimensional neighborhood matrix is center data as a result, with the 32nd to the 34th row
Pixel data instance, obtained calculated result corresponds to each of the 33rd row pixel.
Carrying out operation to the data in table 1, the results are shown in Table 2.X indicate it is unknown, be due to according to data in table 1 simultaneously
It cannot obtain the operation result of corresponding pixel points.
Table 2
| 33rd row | … | x | 2 | 4 | 4 | 0 | x | … |
Edge feature refers to that the point is different from the value put around it, the three-dimensional matrice of first output are as follows:
Obviously, central point 1, value is also 1 to corresponding same column up and down, that is, illustrates that central point does not have vertical boundary
Feature；Similarly, horizontal direction, the left side 0, the right 1, central point is different from left side point, that is, illustrates that the point has horizontal boundary
Feature.So generally speaking, which has boundary characteristic.And passing through convolution algorithm, corresponding operation result is 2, is not equal to
Edge preset threshold value 0, so, the edge feature value of the point is assigned a value of 0；
Equally, second matrix are as follows:
Obviously, central point 1 does not have edge feature in vertical direction, but has edge special in the horizontal direction
Sign so the point is the point with edge feature, and passes through convolution algorithm, and obtaining the point processing result is 4, pre- not equal to edge
If threshold value 0, so, the edge feature value of the point is assigned a value of 0, and so on, obtain final edge feature value such as 3 institute of table
Show.
Table 3
| 33rd row | … | 0 | 0 | 0 | 1 | … |
As long as indicating that the point has boundary characteristic that is, the result that convolution algorithm obtains not is 0, needing to assign
Value is 0, indicates that the point has boundary characteristic；Convolution results are 0, indicate do not have boundary characteristic, are assigned a value of 1, indicate that the point is
Background dot.
By above formula analysis it is found that the implementation of current merging treatment makes the convolution operation of input data and operator exist
Many expenses are saved above resource, and obtained convolution results are identical, are that treatment effect is consistent.Compared to traditional
Convolution algorithm is distinguished to both direction operator, this implementation can save the expense of 5 adders, save resource utilization and reach
55.5%.
It is random to select one in order to intuitively verify the effect of the bianry image edge extraction device 100 in the present embodiment
It is special to generate edge as shown in figure 5, being handled using the bianry image edge extraction device 100 in the present embodiment for bianry image
It is as shown in Figure 6 to levy image.
It, can be by the side of subtlety in image by comparing original bianry image and the edge feature image by extracting
Edge feature is all extracted, and is integrated by the hardware module to entire extraction element, and the dominant frequency of the device can
Reach 140MHz, so the extraction element is an advantage over existing processing equipment in the speed of service and treatment effect, and meets real-time
Property processing requirement.
The technical solution for having been described in detail above with reference to the accompanying drawings the application, present applicant proposes a kind of bianry image edges to mention
Take device, comprising: acquiring unit, storage unit, processing unit and image output unit；Acquiring unit is to be identified for obtaining
The bianry image of image, and according to the rank addresses of pixel in bianry image, the point-by-point pixel number evidence for obtaining pixel；It deposits
Storage unit is used to according to three dimensional neighborhood matrix form, store pixel number evidence, and export to processing unit according to rank addresses；
Processing unit is used for according to rank addresses, to receive, three dimensional neighborhood matrix form, pixel number according to progress edge feature
Operation is extracted, the result of Edge Gradient Feature operation is denoted as edge feature value, and edge feature value is input to image output
Unit；Image output unit is used to generate and export the edge feature image of images to be recognized.Pass through the technical side in the application
Case reduces the complexity of edge extracting operation, saves data space.
Step in the application can be sequentially adjusted, combined, and deleted according to actual needs.
Unit in the application device can be combined, divided and deleted according to actual needs.
Although disclosing the application in detail with reference to attached drawing, it will be appreciated that, these descriptions are only exemplary, not
For limiting the application of the application.The protection scope of the application may include not departing from this Shen by appended claims
It please be in the case where protection scope and spirit for various modifications, remodeling and equivalent scheme made by inventing.
Claims (6)
1. a kind of bianry image edge extraction device, which is characterized in that the device includes: acquiring unit, storage unit, processing list
Member and image output unit；
The acquiring unit is used to obtain the bianry image of images to be recognized, and according to the ranks of pixel in the bianry image
Address obtains the pixel number evidence of the pixel point by point；
The storage unit is used to store the pixel number evidence according to three dimensional neighborhood matrix form according to the rank addresses,
And it exports to the processing unit；
The processing unit is used for according to the rank addresses, to the pixel receive, three dimensional neighborhood matrix form, described
Data carry out Edge Gradient Feature operation, and the result of Edge Gradient Feature operation is denoted as edge feature value, and by the edge
Characteristic value is input to described image output unit；
Described image output unit is used to generate and export the edge feature image of the images to be recognized.
2. bianry image edge extraction device as described in claim 1, which is characterized in that the acquiring unit includes: ranks
Counter and judging unit；
The line count device is set to the input terminal of the acquiring unit, and the line count device is used for described in getting
The pixel in bianry image is counted, and generates the rank addresses, wherein the rank addresses include column address
And row address；
The judging unit is used to be equal to the first preset threshold or the row address when the column address for determining the pixel
When equal to the second preset threshold, the pixel number evidence of the pixel is delivered to described image output unit.
3. bianry image edge extraction device as described in claim 1, which is characterized in that the storage unit is deposited including first
Store up module, the second memory module and level-one register, wherein first memory module and second memory module is first
Into first go out type memory；
The data input pin of the storage unit is connected to input terminal and the level-one deposit of first memory module
The input terminal of device；
The output end of the level-one register is connected to the third output end of the storage unit；
The output end of first memory module is connected to the second output terminal and second storage of the storage unit
The input terminal of module；
The output end of second memory module is connected to the first output end of the storage unit.
4. bianry image edge extraction device as claimed in claim 2, which is characterized in that the processing unit includes: data
Extraction unit and arithmetic element；
The data extracting unit is used to be all larger than or be equal to third predetermined threshold value when the judgement column address and the row address
When, it extracts and sends the pixel number of three dimensional neighborhood matrix form according to the arithmetic element；
The arithmetic element is used to generate the edge feature according to Edge Gradient Feature operation is carried out according to the pixel number
Value.
5. bianry image edge extraction device as claimed in claim 4, which is characterized in that the arithmetic element further include: the
One adder, second adder, third adder, the first shift unit, the second shift unit and comparing unit；
The first adder and the second adder are arranged parallel, the number of the first adder and the second adder
According to output end, pass through first shift unit and second shift unit, the number for being connected to the third adder respectively
According to input terminal, the data output end of the third adder is connected to the comparing unit；
The comparing unit is used for according to the size between the operation result and the edge preset threshold value of the third adder,
Edge Gradient Feature is carried out, the edge feature value is exported.
6. bianry image edge extraction device as claimed in claim 5, which is characterized in that extract and send three dimensional neighborhood matrix
Pixel number described in form is specifically included according to the arithmetic element:
Data decimation frame is generated, the data decimation frame is 3 × 3 matrix forms；
The pixel number evidence that will be exported according to three dimensional neighborhood matrix form, is delivered to the data decimation frame；
Successively extract the first row first row of the data decimation frame, the pixel of the first row secondary series, the second row first row
Point data is delivered to the first adder；
Successively extract the second row third column, the third line secondary series, the tertial pixel of the third line of the data decimation frame
Point data is delivered to the second adder.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910535278.1A CN110264488B (en) | 2019-06-20 | 2019-06-20 | Binary image edge extraction device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910535278.1A CN110264488B (en) | 2019-06-20 | 2019-06-20 | Binary image edge extraction device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110264488A true CN110264488A (en) | 2019-09-20 |
| CN110264488B CN110264488B (en) | 2021-03-16 |
Family
ID=67919679
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910535278.1A Expired - Fee Related CN110264488B (en) | 2019-06-20 | 2019-06-20 | Binary image edge extraction device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110264488B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113068045A (en) * | 2021-03-17 | 2021-07-02 | 厦门雅基软件有限公司 | Data storage method and device, electronic equipment and computer readable storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003105087A1 (en) * | 2002-06-10 | 2003-12-18 | Lockeed Martin Corporation | Edge detection using hough transformation |
| CN106485255A (en) * | 2016-09-29 | 2017-03-08 | 深圳元启智能技术有限公司 | A kind of DM code positioning and the method and system of identification |
| US20170301066A1 (en) * | 2015-10-14 | 2017-10-19 | Shanghai United Imaging Healthcare Co., Ltd. | System and method for image correction |
| WO2017193876A1 (en) * | 2016-05-12 | 2017-11-16 | 深圳市太赫兹科技创新研究院 | Method and device for detecting dangerous object hidden on human body from microwave image |
| CN108537786A (en) * | 2018-03-30 | 2018-09-14 | 百度在线网络技术（北京）有限公司 | For handling image method and device |
| CN108716890A (en) * | 2018-08-17 | 2018-10-30 | 苏州富鑫林光电科技有限公司 | A kind of high-precision size detecting method based on machine vision |
| CN109427066A (en) * | 2017-08-31 | 2019-03-05 | 中国科学院微电子研究所 | Edge detection method for any angle |
-
2019
- 2019-06-20 CN CN201910535278.1A patent/CN110264488B/en not_active Expired - Fee Related
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003105087A1 (en) * | 2002-06-10 | 2003-12-18 | Lockeed Martin Corporation | Edge detection using hough transformation |
| US20170301066A1 (en) * | 2015-10-14 | 2017-10-19 | Shanghai United Imaging Healthcare Co., Ltd. | System and method for image correction |
| WO2017193876A1 (en) * | 2016-05-12 | 2017-11-16 | 深圳市太赫兹科技创新研究院 | Method and device for detecting dangerous object hidden on human body from microwave image |
| CN106485255A (en) * | 2016-09-29 | 2017-03-08 | 深圳元启智能技术有限公司 | A kind of DM code positioning and the method and system of identification |
| CN109427066A (en) * | 2017-08-31 | 2019-03-05 | 中国科学院微电子研究所 | Edge detection method for any angle |
| CN108537786A (en) * | 2018-03-30 | 2018-09-14 | 百度在线网络技术（北京）有限公司 | For handling image method and device |
| CN108716890A (en) * | 2018-08-17 | 2018-10-30 | 苏州富鑫林光电科技有限公司 | A kind of high-precision size detecting method based on machine vision |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113068045A (en) * | 2021-03-17 | 2021-07-02 | 厦门雅基软件有限公司 | Data storage method and device, electronic equipment and computer readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110264488B (en) | 2021-03-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107301402B (en) | Method, device, medium and equipment for determining key frame of real scene | |
| CN108596944B (en) | Method and device for extracting moving target and terminal equipment | |
| KR101639852B1 (en) | Pixel value compaction for graphics processing | |
| CN107025660B (en) | Method and device for determining image parallax of binocular dynamic vision sensor | |
| US20140192158A1 (en) | Stereo Image Matching | |
| TWI775935B (en) | Method of estimating depth in image | |
| Ttofis et al. | Towards accurate hardware stereo correspondence: A real-time fpga implementation of a segmentation-based adaptive support weight algorithm | |
| CN104282027B (en) | Circle detecting method based on Hough transformation | |
| JP2017535904A (en) | System and method for object detection | |
| CN111192312B (en) | Depth image acquisition method, device, equipment and medium based on deep learning | |
| CN116310098A (en) | Multi-view three-dimensional reconstruction method based on attention mechanism and variable convolution depth network | |
| CN109741306B (en) | Image processing method applied to dangerous chemical storehouse stacking | |
| CN114943659A (en) | Map area segmentation method and related device | |
| CN110264488A (en) | A kind of bianry image edge extraction device | |
| Park et al. | A vision processor with a unified interest-point detection and matching hardware for accelerating a stereo-matching algorithm | |
| CN109949204A (en) | The asterism mass center of pipeline organization extracts circuit | |
| US10824877B2 (en) | Hierarchical data organization for dense optical flow processing in a computer vision system | |
| CN115620321B (en) | Table identification method and device, electronic equipment and storage medium | |
| CN116403200A (en) | License plate real-time identification system based on hardware acceleration | |
| CN106971386A (en) | Judge method, device and the client device of image integrity degree and page loading degree | |
| CN111242140B (en) | Method for rapidly extracting contour line under non-uniform illumination | |
| RU2383925C2 (en) | Method of detecting contours of image objects and device for realising said method | |
| WO2019000300A1 (en) | Two-dimensional data matching method, device and logic circuit | |
| CN112749713A (en) | Big data image recognition system and method based on artificial intelligence | |
| CN112967321A (en) | Moving object detection method and device, terminal equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210316 |