CN105844608A - Urinary sediment image segmentation method and urinary sediment image segmentation device - Google Patents
Urinary sediment image segmentation method and urinary sediment image segmentation device Download PDFInfo
- Publication number
- CN105844608A CN105844608A CN201510024371.8A CN201510024371A CN105844608A CN 105844608 A CN105844608 A CN 105844608A CN 201510024371 A CN201510024371 A CN 201510024371A CN 105844608 A CN105844608 A CN 105844608A
- Authority
- CN
- China
- Prior art keywords
- gray
- threshold
- image
- value
- gray scale
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Facsimile Image Signal Circuits (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides a urinary sediment image segmentation method and a urinary sediment image segmentation device. The method comprises the following steps: calculating the gray histogram of an image to be segmented; judging whether the gray histogram is a single-peak histogram; if the gray histogram is a single-peak histogram, looking for the gray lower limit value and the gray upper limit value of the single peak in the gray histogram; and segmenting the foreground and background of the image according to the gray lower limit value and the gray upper limit value, wherein the pixels of which the gray values are between the gray lower limit value and the gray upper limit value in the image are segmented as the background, and the other pixels in the image are segmented as the foreground. Through the method and the device, the foreground and background of a urinary sediment image can be segmented quickly, the computation complexity is low, and the amount of computation is small.
Description
Technical field
The present invention relates to image processing field, particularly relate to dividing method and the dress of a kind of urine sediment image
Put.
Background technology
Sediment urinalysis analysis is a kind of Noninvasive testing, for detecting the various visible components in urine,
Such as, the number of erythrocyte, leukocyte, cast (CAST), epithelial cell, crystallization etc..
In traditional sediment urinalysis analytical technology, shoot urine specimen figure first with micro imaging system
Picture.Then, by the way of cervical arthroplasty (MME), the visible component in urine sediment image is counted.
Cervical arthroplasty needs manually to count, and therefore would generally expend longer time, and the mirror of different observers
Inspection result there may be difference.
Along with the fast development of computer science, digital medical images technology be increasingly becoming one important auxiliary
Help detection means.By computer technology urine sediment image processed and analyze, it is possible to automatically will
Visible component in image splits, and classifies various visible components and identify, thus part
Or substitute cervical arthroplasty fully, accelerate the speed of detection, improve the concordance of testing result.
Image segmentation be urine sediment image process in a committed step, for by display foreground (if any
Formation grades area-of-interest) split from background, it is classification and the identification of urinary formed element
Provide important foundation.The image that result is binaryzation of image segmentation, wherein, value is the pixel of 1
Expression prospect, value is the pixels representing background of 0.
The image partition method that presently, there are such as has adaboost (Adaptive Boosting) classification to calculate
Method and saliency detection (Saliency Detection) algorithm.Adaboost algorithm needs training point
Class device, and calculate with entire image, it is therefore desirable to the longer training time, and there is bigger fortune
Calculate complexity.Saliency detection algorithm needs to calculate the Local textural feature of image, and needs to carry
Take Analysis On Multi-scale Features, be therefore also required to bigger operand.
Summary of the invention
In view of this, it is an object of the invention to provide dividing method and the device of a kind of urine sediment image,
Can split the foreground and background of urine sediment image rapidly, and computational complexity is low, and operand is little.
According to one embodiment of present invention, it is provided that the dividing method of a kind of urine sediment image, including:
Calculate the grey level histogram of an image to be split;
Judge whether described grey level histogram is unimodal histogram;
If unimodal histogram, then in described grey level histogram, find described unimodal gray scale lower limit
With gray scale higher limit;
According to described gray scale lower limit and the foreground and background of the gray scale higher limit described image of segmentation, wherein,
The gray value in described image pixel between described gray scale lower limit and gray scale higher limit is divided
For background, the rest of pixels in described image is divided into prospect.
According to another embodiment of the invention, it is provided that the segmenting device of a kind of urine sediment image, including:
Computing module, for calculating the grey level histogram of an image to be split;
Judge module, is used for judging whether described grey level histogram is unimodal histogram;
Search module, for when the judged result of described judge module is unimodal histogram, at described ash
Degree rectangular histogram finds described unimodal gray scale lower limit and gray scale higher limit;
Segmentation module, for splitting the prospect of described image according to described gray scale lower limit and gray scale higher limit
And background, wherein, the gray value in described image is between described gray scale lower limit and gray scale higher limit
Pixel be divided into background, the rest of pixels in described image is divided into prospect.
By according to embodiments of the invention, the grey level histogram based on the urine sediment image prospect to image
Split with background, it is not necessary to calculate complicated textural characteristics and prior long-time training, and
Having only to process on single yardstick, therefore detection speed is fast, and computational complexity is low, and operand is little.
Accompanying drawing explanation
Below with reference to accompanying drawing, by coming according to a particular embodiment of the invention the purpose of the present invention, spy
Effect of seeking peace is described in detail.These explanations are merely cited for, not in order to limit the protection of the present invention
Scope.Wherein:
Fig. 1 shows the method flow schematic diagram of according to embodiments of the present invention;
Fig. 2 shows the schematic flow sheet of the embodiment two according to the inventive method;
Fig. 3 shows and sets the first gray threshold and the second ash in the embodiment three according to the inventive method
The schematic diagram of degree threshold value;
Fig. 4 shows the schematic diagram of setpoint frequency threshold value in the embodiment four according to the inventive method;
Fig. 5 shows the schematic diagram of setpoint frequency threshold value in the embodiment five according to the inventive method;
Fig. 6 shows and compensates the first gray threshold and the second ash in the embodiment six according to the inventive method
The schematic diagram of degree threshold value;
Fig. 7 shows the schematic diagram of the segmenting device of the urine sediment image of according to embodiments of the present invention seven;
Fig. 8 shows the schematic diagram of search module in the embodiment eight according to apparatus of the present invention;
Fig. 9 shows according to the schematic diagram of gray threshold setting module in the embodiment of apparatus of the present invention;
Figure 10 shows the schematic diagram of the embodiment medium frequency threshold value setting module according to apparatus of the present invention;
Figure 11 shows the signal of the another embodiment medium frequency threshold value setting module according to apparatus of the present invention
Figure;
Figure 12 shows the schematic diagram of the splitting equipment of urine sediment image according to embodiments of the present invention.
Detailed description of the invention
Fig. 1 shows the method flow schematic diagram of according to embodiments of the present invention.As it is shown in figure 1, at this
In embodiment, the dividing method of urine sediment image comprises the following steps:
S2: calculate the grey level histogram of an image to be split.
Grey level histogram (histogram) is the function of gray value, and its abscissa is gray value, vertical coordinate
It it is the frequency of this gray value appearance.The grey level histogram of image represents in this image have every kind of gray value
The number of pixel, has the frequency that the pixel of every kind of gray value occurs, is the one of image in reflection image
Statistical nature.
The calculating of grey level histogram is the simplest.If image has L level gray scale, then size is the ash of M*N
Degree image f (x, grey level histogram hist y) [0,1 ..., L-1] such as can be by acquisition calculated as below:
A, initialization hist [k]=0;K=0,1 ..., L-1
B, statistics hist [f (x, y)] ++;X, y=0 ..., M-1,0 ..., N-1
C, standardization hist [f (x, y)] /=M*N
If image to be split is RGB image, it is possible to use RGB is schemed by known or that other is suitable method
As transferring gray level image to, then calculate the grey level histogram of this gray level image.
S4: judge whether described grey level histogram is unimodal histogram.
If most energy of grey level histogram concentrate near maximal peak point, then it is believed that be somebody's turn to do
Grey level histogram is unimodal histogram.
Judge that whether grey level histogram is that the specific threshold of unimodal histogram can be according to specifically applying according to warp
Test value or determined by emulation and experiment.Such as, if the energy of at least the 80% of grey level histogram is concentrated
In the intensity value ranges of maximal peak point both sides at most ± 2.5%, then it is believed that this grey level histogram is
Unimodal histogram.
S6: if unimodal histogram, then find under described unimodal gray scale in described grey level histogram
Limit value and gray scale higher limit.
If grey level histogram is unimodal histogram, then show that in image, first kind pixel accounts for the overwhelming majority,
And the gray value of first kind pixel is all in the range of a concentration;And Equations of The Second Kind pixel accounts for very few, the
The gray value of two class pixels can be distinguished significantly with the gray value of first kind pixel.
Therefore, it can in grey level histogram, find unimodal gray scale lower limit and gray scale higher limit, and profit
By this gray scale lower limit and gray scale higher limit to first kind pixel and Equations of The Second Kind pixel classifications.
In first kind pixel and Equations of The Second Kind pixel, a usual class is background pixel, another kind of for prospect picture
Element, then can realize the segmentation of foreground and background by classification.
S8: split the foreground and background of described image according to described gray scale lower limit and gray scale higher limit, its
In, the pixel between described gray scale lower limit and gray scale higher limit of the gray value in described image is divided
Being segmented into background, the rest of pixels in described image is divided into prospect.
The probability that visible component in urine occurs in urine specimen is low, quantity is few, therefore, heavy at urine
In slag image, background pixel accounts for the overwhelming majority, and foreground pixel accounts for very few.The then gray value in image
Pixel between described gray scale lower limit and gray scale higher limit is divided into background, in described image
Rest of pixels is divided into prospect.
As can be seen here, in an embodiment according to the present invention, grey level histogram based on urine sediment image,
According to gray scale lower limit unimodal in this grey level histogram and gray scale higher limit, just can to the prospect of image and
Background is split.Notable compared to adaboost (Adaptive Boosting) sorting algorithm and image
Property detection (Saliency Detection) algorithm, need not to calculate complicated according to embodiments of the invention
Textural characteristics and prior long-time training, and have only to process on single yardstick, therefore detect
Speed is fast, and computational complexity is low, and operand is little.
It was found by the inventors of the present invention that the grey level histogram of urine sediment image is usually unimodal histogram,
The probability that visible component in this explanation urine occurs in urine specimen is low, quantity is few, accordingly,
In urine sediment image, major part image is background image, and the foreground image with area-of-interest only accounts for very
A little part.Therefore, the grey level histogram utilizing urine sediment image is usually unimodal histogram, and this is special
Property, can effectively split the foreground and background of image, simple and practical.
In addition, it is contemplated that the most of image in urine sediment image is background image, compared to cervical arthroplasty
Mode need manually to find foreground image and visible component counted, utilize the reality according to the present invention
Execute example, can fast and effeciently filter out the foreground image in urine sediment image, save the detection time, and
And avoid the discordance of cervical arthroplasty result, for the classification of urinary formed element with identify and provide good
Basis.
Below will be by the present invention is described in more detail according to a preferred embodiment of the invention.
Fig. 2 shows the schematic flow sheet of the embodiment two according to the inventive method.As in figure 2 it is shown,
In this embodiment two, the dividing method of urine sediment image comprises the following steps:
S20: calculate the grey level histogram of an image to be split.
If image f to be split (x, y) has L level gray scale (such as L=256), and pixel count is M*N, its
Grey level histogram hist [0,1 ..., L-1] by acquisition calculated as below:
A, initialization hist [k]=0;K=0,1 ..., L-1
B, statistics hist [f (x, y)] ++;X, y=0 ..., M-1,0 ..., N-1
C, standardization hist [f (x, y)] /=M*N
S40: judge whether described grey level histogram is unimodal histogram.
In described grey level histogram, search out maximal peak point, namely the overall situation of grey level histogram is maximum
Value.If the energy of the 80% of described grey level histogram concentrates on the gray value model of maximal peak point both sides ± 5
In enclosing, then it is assumed that this grey level histogram is unimodal histogram.
S60: if unimodal histogram, it is determined that the gray scale of maximal peak point in described grey level histogram
Value.
When described grey level histogram is unimodal histogram, record the gray scale of described maximal peak point
Value.
S62: start to the first default gray threshold from the gray value of described maximal peak point, by
One judges that the frequency of current grayvalue, whether less than a default frequency threshold, is preset if less than described
Frequency threshold, then current grayvalue is set as described gray scale lower limit, otherwise, by described first ash
Degree threshold value is set as described gray scale lower limit;
Start to the second default gray threshold from the gray value of described maximal peak point, judge one by one
Whether the frequency of current grayvalue is less than described default frequency threshold, if less than described default frequency
Threshold value, then be set as described gray scale higher limit, otherwise, by described second gray threshold by current grayvalue
It is set as described gray scale higher limit.
Described first gray threshold, the second gray threshold and frequency threshold can according to concrete application empirically
Value or by emulation and experiment determine.By above-mentioned steps S60 and S62, just can be straight in described gray scale
When side's figure is unimodal histogram, determine described unimodal gray scale lower limit and gray scale higher limit.In step
In S62, set the operation of described gray scale higher limit and gray scale lower limit regardless of front and back, it is possible to carry out parallel.
S80: split the foreground and background of described image according to described gray scale lower limit and gray scale higher limit,
Wherein, the pixel quilt between described gray scale lower limit and gray scale higher limit of the gray value in described image
Being divided into background, the rest of pixels in described image is divided into prospect.
So far, just can be based on described grey level histogram, simply and effectively by described image to be split
Foreground and background is separated.
In order to improve detection speed further, reduce computational complexity, as in figure 2 it is shown, in embodiment two
In, also can farther include step S10: to described image drop sampling to be split.Using, this is down-sampled
When step, described in above-mentioned steps S20 calculate grey level histogram be: calculate described down-sampled after
The grey level histogram of image.By to image drop sampling to be split, it is possible to reduce the pixel count in image,
Thus reduce the operand calculating grey level histogram, and the characteristic of grey level histogram will not be changed.
Below in conjunction with Fig. 3 to Fig. 5 for presetting described first gray threshold, the second gray threshold and frequency
The operation of rate threshold value is illustrated.
Fig. 3 shows the schematic flow sheet of the embodiment three according to the inventive method.As it is shown on figure 3,
In this embodiment three, the operation presetting described first gray threshold and the second gray threshold includes:
S620: obtain the grey level histogram of several urine sediment images having split foreground and background.
In this step, several undivided images are randomly selected.For the image selected by every width, calculate
Its grey level histogram, and manually regulate or automatically regulate for splitting foreground and background to set interval
The upper limit value and lower limit value of gray value;Under the conditions of different gray values, observe the prospect being partitioned into;When dividing
When area-of-interest in the prospect cut out and image coincide (area-of-interest of for example, at least 90% by
Split), record lower limit and the higher limit of gray value under the conditions of this.
S621: determine described several minima of gray value of background pixel having split image and maximums
Value.
For each image, the lower limit of the gray value recorded and higher limit are the gray scale of background pixel
The lower limit of value and higher limit.The relatively lower limit of the gray value of the background pixel of described multiple image, really
The minima of a fixed gray value;The relatively higher limit of the gray value of the background pixel of described multiple image,
Determine the maximum of a gray value.
S622: using the minima of described gray value as described first gray threshold.
S623: using the maximum of described gray value as described second gray threshold.
By the operation in this embodiment three, just described first gray threshold can be determined by the way of statistics
With the second gray threshold, and determined by the first gray threshold and the second gray threshold can meet concrete application
Needs.
In above-mentioned steps S620, several urine sediment images having split foreground and background described can also
It is the urine sediment image having split foreground and background by the way of artificial mark, or otherwise
Split the urine sediment image of foreground and background.
Fig. 4 shows the schematic flow sheet of the embodiment four according to the inventive method.As shown in Figure 4, exist
In this embodiment four, the operation presetting described frequency threshold includes:
S624: obtain the grey level histogram of several urine sediment images having split foreground and background.
In this step, several undivided images are randomly selected.For the image selected by every width, calculate
Its grey level histogram, and manually regulate or automatically regulate for splitting foreground and background to set interval
The upper limit value and lower limit value of gray value;Under the conditions of different gray values, observe the prospect being partitioned into;When dividing
When area-of-interest in the prospect cut out and image coincide (area-of-interest of for example, at least 90% by
Split), record the lower-frequency limit value of background pixel under the conditions of this, namely the gray scale under the conditions of this
The frequency values that the frequency values corresponding to lower limit of value is corresponding with the higher limit of gray value.
S625: determine that described several have split the average of lower-frequency limit value of background pixel of image;
In this step, the lower-frequency limit value of the background pixel of described multiple image is carried out statistical average.
S626: using described average as described frequency threshold.
By the assembly average of above-mentioned steps gained i.e. by as described frequency threshold.
In embodiment four, again by the mode of statistics, just can determine institute according to concrete application scenarios
State frequency threshold.
Fig. 5 shows the schematic flow sheet of the embodiment five according to the inventive method.As it is shown in figure 5,
In this embodiment five, the another kind of operation presetting described frequency threshold includes:
S627: determine the pixel count of described image to be split and the ratio of number of greyscale levels.
In this step, if (x, y) has L level gray scale to image f to be split, and pixel count is M*N, then
Calculate M*N/L.
If image to be split has been carried out down-sampled, calculate the most in this step described down-sampled after figure
The pixel count of picture and the ratio of number of greyscale levels.
S628: using described ratio as described frequency threshold.
In this embodiment five, set frequency threshold and the pixel count positive correlation of image to be split, accord with
Close the characteristic of unimodal histogram, and simple.
In a particular application, because image-forming block is aging or the impact of the factor such as ambient lighting, it is likely to result in and treats
The fluctuation of the gray value of segmentation image.If the first default gray threshold and the second gray threshold are carried out
Adjustment, then can better compensate for the fluctuation of the gray value of image to be split.
Fig. 6 shows the schematic flow sheet of the embodiment six according to the inventive method.As shown in Figure 6, exist
In this embodiment six, the operation of adjustment the first gray threshold and the second gray threshold includes:
S61: judge whether the gray value of described maximal peak point is positioned at described first gray threshold and described
Between second gray threshold.
Such as, by step S60 of above-mentioned embodiment illustrated in fig. 2 one, it has been determined that described maximum peak
The gray value of value point.The most in this step, the gray value whether position of described maximal peak point is determined whether
Between described first gray threshold and described second gray threshold.
S63: if located between described first gray threshold and described second gray threshold, the most directly hold
The described setting gray scale lower limit of row and the step of described gray scale higher limit;
Otherwise, according to intermediate value and the described maximum peak of described first gray threshold and described second gray threshold
The difference of the gray value of value point, compensates described first gray threshold and described second gray threshold, and uses benefit
The first gray threshold after repaying and compensate after second gray threshold perform described setting gray scale lower limit and
The step of described gray scale higher limit.
In this step, if above-mentioned judged result is for certainly, the most directly performing such as embodiment illustrated in fig. 2
Step S62 of one;Otherwise, first described first gray threshold and described second gray threshold are compensated,
Perform step S62 of such as embodiment illustrated in fig. 2 one the most again.So, just can better compensate for treating point
Cut the fluctuation of the gray value of image.
Above in association with accompanying drawing, the multiple embodiments according to the inventive method are illustrated respectively, this area
Those of skill will appreciate that, in a particular application, the various embodiments described above are carried out appropriately also dependent on needs
Ground combines.
Corresponding to the method for the present invention, additionally provide a kind of urine sediment image according to embodiments of the invention seven
Segmenting device, as it is shown in fig. 7, this segmenting device 70 includes:
Computing module 72, for calculating the grey level histogram of an image to be split;
Judge module 74, is used for judging whether described grey level histogram is unimodal histogram;
Search module 76, for when the judged result of described judge module is unimodal histogram, in institute
State and grey level histogram is found described unimodal gray scale lower limit and gray scale higher limit;And
Segmentation module 78, for splitting described image according to described gray scale lower limit and gray scale higher limit
Foreground and background, wherein, the gray value in described image is positioned at described gray scale lower limit and gray scale higher limit
Between pixel be divided into background, the rest of pixels in described image is divided into prospect.
Utilizing the segmenting device 70 of this embodiment seven, grey level histogram based on urine sediment image is to image
Foreground and background split, it is not necessary to calculate complicated textural characteristics and prior long-time instruction
Practicing, and have only to process on single yardstick, therefore detection speed is fast, and computational complexity is low, computing
Measure little.
In order to improve detection speed further, reduce computational complexity, as it is shown in fig. 7, fill in this segmentation
Put in 70, also can farther include down-sampled module 71, for described image drop sampling to be split.
When have employed these down-sampled module 71, above-mentioned computing module 72 is then used for calculating described fall and adopts
The grey level histogram of the image after sample.By to image drop sampling to be split, it is possible to reduce the picture in image
Prime number, thus reduce the operand calculating grey level histogram, and the characteristic of grey level histogram will not be changed.
Fig. 8 shows the structural representation of the embodiment eight according to apparatus of the present invention.As shown in Figure 8, exist
In this embodiment eight, search module 86 includes:
First determines unit 862, for determining the gray value of maximal peak point in described grey level histogram;
Lower limit setup unit 864, for starting to default the from the gray value of described maximal peak point
Till one gray threshold, judge that whether the frequency of current grayvalue is less than a default frequency threshold one by one
Value, if less than described default frequency threshold, is then set as described gray scale lower limit by current grayvalue,
Otherwise, described first gray threshold is set as described gray scale lower limit;And
Upper limit value setting unit 866, for starting to default the from the gray value of described maximal peak point
Till two gray thresholds, judge that whether the frequency of current grayvalue is less than described default frequency threshold one by one
Value, if less than described default frequency threshold, is then set as described gray scale higher limit by current grayvalue,
Otherwise, described second gray threshold is set as described gray scale higher limit.
In order to better compensate for the fluctuation of the gray value of image to be split, eliminating because image-forming block is aging or
The impact of the factors such as ambient lighting, as shown in Figure 8, this search module 86 also can farther include:
Judging unit 863, for judging whether the gray value of described maximal peak point is positioned at described first ash
Between degree threshold value and described second gray threshold;And
Compensating unit 865, for when the judged result of described judging unit is no, according to described first
The intermediate value of gray threshold and described second gray threshold and the difference of the gray value of described maximal peak point, compensate
Described first gray threshold and described second gray threshold, and will compensate after the first gray threshold as in advance
If the first gray threshold, will compensate after the second gray threshold as the second default gray threshold.
Described first gray threshold, the second gray threshold and frequency threshold can according to concrete application empirically
Value or by emulation and experiment determine.Below in conjunction with Fig. 9 to Figure 11 for setting described first ash
The module of degree threshold value, the second gray threshold and frequency threshold is illustrated.
As it is shown in figure 9, according in the embodiment of apparatus of the present invention, may also include gray threshold and set mould
Block 73, this gray threshold setting module 73 includes:
Acquiring unit 730 is straight for obtaining the gray scale of several urine sediment images having split foreground and background
Fang Tu;
Second determines unit 732, for determining that described several have split the gray value of background pixel of image
Minima and maximum;
First setup unit 734, is used for the minima of described gray value as described first gray threshold;
Second setup unit 736, is used for the maximum of described gray value as described second gray threshold.
As shown in Figure 10, according in another embodiment of apparatus of the present invention, may also include frequency threshold
Setting module 75, this frequency threshold setting module 75 includes:
Acquiring unit 752, obtains the grey level histogram of several urine sediment images having split foreground and background;
3rd determines unit 754, under the frequency determining several background pixels having split image described
The average of limit value;
3rd setup unit 756, is used for described average as described frequency threshold.
By the gray threshold setting module 73 in above-described embodiment and frequency threshold setting module 75, just
Can by emulation and experiment use statistics by the way of set described first gray threshold, the second gray threshold and
Frequency threshold.
Figure 11 shows and sets mould according to frequency threshold included in the another embodiment of apparatus of the present invention
Block 77, this frequency threshold setting module 77 includes:
4th determines unit 772, for determining the pixel count of described image to be split and the ratio of number of greyscale levels
Value;
4th setup unit 774, is used for described ratio as described frequency threshold.
By the frequency threshold setting module 77 in this embodiment, set frequency threshold is with to be split
The pixel count positive correlation of image, meets the characteristic of unimodal histogram, and simple.
It will be understood to those skilled in the art that above-mentioned each embodiment according to apparatus of the present invention can utilize
Software, hardware (such as integrated circuit, DSP or FPGA etc.) or the mode of software and hardware combining realize,
And can combine the most rightly.
The splitting equipment of a kind of urine sediment image is additionally provided, such as Figure 12 institute according to embodiments of the invention
Showing, this splitting equipment 120 includes: memorizer 122 and processor 124.Memorizer 122 is used for storing
Executable instruction.Processor 124, for the executable instruction stored according to described memorizer 122, is held
The row step included by method according to embodiments of the present invention.
Detailed description for the embodiment according to this transmitting apparatus and equipment is referred to according to side of the present invention
The relevant explanation of the embodiment of method, does not repeats them here.
Additionally, also provide for a kind of machine readable media according to one embodiment of present invention, on it, storage has
Executable instruction, when this executable instruction is performed so that machine performs performed by processor 124
Operation.
It will be appreciated by those skilled in the art that each embodiment above can be without departing from invention essence
In the case of make various changes and modifications, therefore, protection scope of the present invention should be by appended right
Claim limits.
Claims (16)
1. the dividing method of urine sediment image, including:
Calculate the grey level histogram of an image to be split;
Judge whether described grey level histogram is unimodal histogram;
If unimodal histogram, then in described grey level histogram, find described unimodal gray scale lower limit
With gray scale higher limit;
According to described gray scale lower limit and the foreground and background of the gray scale higher limit described image of segmentation, wherein,
The gray value in described image pixel between described gray scale lower limit and gray scale higher limit is divided
For background, the rest of pixels in described image is divided into prospect.
2. the method for claim 1, it is characterised in that the described unimodal gray scale of described searching
Lower limit and gray scale higher limit include:
Determine the gray value of maximal peak point in described grey level histogram;
Start to the first default gray threshold from the gray value of described maximal peak point, judge one by one
Whether the frequency of current grayvalue is less than a default frequency threshold, if less than described default frequency
Threshold value, then be set as described gray scale lower limit, otherwise, by described first gray threshold by current grayvalue
It is set as described gray scale lower limit;
Start to the second default gray threshold from the gray value of described maximal peak point, judge one by one
Whether the frequency of current grayvalue is less than described default frequency threshold, if less than described default frequency
Threshold value, then be set as described gray scale higher limit, otherwise, by described second gray threshold by current grayvalue
It is set as described gray scale higher limit.
3. method as claimed in claim 2, it is characterised in that described first gray threshold and described
Second gray threshold is set by following step:
Obtain the grey level histogram of several urine sediment images having split foreground and background;
Determine described several minima of gray value of background pixel having split image and maximums;
Using the minima of described gray value as described first gray threshold;
Using the maximum of described gray value as described second gray threshold.
4. method as claimed in claim 2, it is characterised in that described frequency threshold is by following step
Rapid setting:
Obtain the grey level histogram of several urine sediment images having split foreground and background;
Determine that described several have split the average of lower-frequency limit value of background pixel of image;
Using described average as described frequency threshold.
5. method as claimed in claim 2, it is characterised in that described frequency threshold is by following step
Rapid setting:
Determine the pixel count of described image to be split and the ratio of number of greyscale levels;
Using described ratio as described frequency threshold.
6. method as claimed in claim 2, it is characterised in that the described unimodal gray scale of described searching
Lower limit and gray scale higher limit also include:
Judge whether the gray value of described maximal peak point is positioned at described first gray threshold and described second
Between gray threshold;
If located between described first gray threshold and described second gray threshold, the most directly perform described
Set gray scale lower limit and the step of described gray scale higher limit;
Otherwise, according to intermediate value and the described maximum peak of described first gray threshold and described second gray threshold
The difference of the gray value of value point, compensates described first gray threshold and described second gray threshold, and uses benefit
The first gray threshold after repaying and compensate after second gray threshold perform described setting gray scale lower limit and
The step of described gray scale higher limit.
7. the method for claim 1, it is characterised in that also include:
To described image drop sampling to be split;
Described calculating grey level histogram is: calculate described down-sampled after the grey level histogram of image.
8. the segmenting device of urine sediment image, including:
Computing module, for calculating the grey level histogram of an image to be split;
Judge module, is used for judging whether described grey level histogram is unimodal histogram;
Search module, for when the judged result of described judge module is unimodal histogram, at described ash
Degree rectangular histogram finds described unimodal gray scale lower limit and gray scale higher limit;
Segmentation module, for splitting the prospect of described image according to described gray scale lower limit and gray scale higher limit
And background, wherein, the gray value in described image is between described gray scale lower limit and gray scale higher limit
Pixel be divided into background, the rest of pixels in described image is divided into prospect.
9. device as claimed in claim 8, it is characterised in that described search module includes:
First determines unit, for determining the gray value of maximal peak point in described grey level histogram;
Lower limit setup unit, for starting to the first default ash from the gray value of described maximal peak point
Till degree threshold value, judge whether the frequency of current grayvalue is less than a default frequency threshold one by one, as
Current grayvalue less than described default frequency threshold, is then set as described gray scale lower limit by fruit, otherwise,
Described first gray threshold is set as described gray scale lower limit;
Upper limit value setting unit, for starting to the second default ash from the gray value of described maximal peak point
Till degree threshold value, judge whether the frequency of current grayvalue is less than described default frequency threshold one by one, as
Current grayvalue less than described default frequency threshold, is then set as described gray scale higher limit by fruit, otherwise,
Described second gray threshold is set as described gray scale higher limit.
10. device as claimed in claim 9, it is characterised in that described device also includes gray threshold
Setting module;Described gray threshold setting module includes:
Acquiring unit, for obtaining the intensity histogram of several urine sediment images having split foreground and background
Figure;
Second determines unit, for determine described several split image background pixel gray value
Little value and maximum;
First setup unit, is used for the minima of described gray value as described first gray threshold;
Second setup unit, is used for the maximum of described gray value as described second gray threshold.
11. devices as claimed in claim 9, it is characterised in that described device also includes frequency threshold
Setting module;Described frequency threshold setting module includes:
Acquiring unit, obtains the grey level histogram of several urine sediment images having split foreground and background;
3rd determines unit, for determining that described several have split the lower-frequency limit value of background pixel of image
Average;
3rd setup unit, is used for described average as described frequency threshold.
12. devices as claimed in claim 9, it is characterised in that described device also includes frequency threshold
Setting module;Described frequency threshold setting module includes:
4th determines unit, for determining the pixel count of described image to be split and the ratio of number of greyscale levels;
4th setup unit, is used for described ratio as described frequency threshold.
13. devices as claimed in claim 9, it is characterised in that described search module also includes:
Judging unit, for judging whether the gray value of described maximal peak point is positioned at described first gray scale threshold
Between value and described second gray threshold;
Compensating unit, for when the judged result of described judging unit is no, according to described first gray scale
The intermediate value of threshold value and described second gray threshold and the difference of the gray value of described maximal peak point, compensate described
First gray threshold and described second gray threshold, and will compensate after the first gray threshold as default
First gray threshold, the second gray threshold after compensating is as the second default gray threshold.
14. devices as claimed in claim 8, it is characterised in that described device also includes:
Down-sampled module, for described image drop sampling to be split;
Described computing module, for calculate described down-sampled after the grey level histogram of image.
The splitting equipment of 15. urine sediment images, including:
Memorizer, is used for storing executable instruction;
Processor, for the executable instruction stored according to described memorizer, performs such as claim
Arbitrary described step included by method in 1-7.
16. 1 kinds of machine readable medias, on it, storage has executable instruction, when described executable instruction quilt
During execution so that machine performs the step included by method as described in arbitrary in claim 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510024371.8A CN105844608A (en) | 2015-01-16 | 2015-01-16 | Urinary sediment image segmentation method and urinary sediment image segmentation device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510024371.8A CN105844608A (en) | 2015-01-16 | 2015-01-16 | Urinary sediment image segmentation method and urinary sediment image segmentation device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105844608A true CN105844608A (en) | 2016-08-10 |
Family
ID=56580314
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510024371.8A Pending CN105844608A (en) | 2015-01-16 | 2015-01-16 | Urinary sediment image segmentation method and urinary sediment image segmentation device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105844608A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110717465A (en) * | 2019-10-15 | 2020-01-21 | 周兴祥 | Identification method and system based on calcium oxalate crystallization |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101101668A (en) * | 2007-07-13 | 2008-01-09 | 杭州电子科技大学 | Medical image cutting method based on oscillatory network |
CN101609504A (en) * | 2009-07-21 | 2009-12-23 | 华中科技大学 | A kind of method for detecting, distinguishing and locating infrared imagery sea-surface target |
CN102289668A (en) * | 2011-09-07 | 2011-12-21 | 谭洪舟 | Binaryzation processing method of self-adaption word image based on pixel neighborhood feature |
CN103310211A (en) * | 2013-04-26 | 2013-09-18 | 四川大学 | Filling mark recognition method based on image processing |
CN103903275A (en) * | 2014-04-23 | 2014-07-02 | 贵州大学 | Method for improving image segmentation effects by using wavelet fusion algorithm |
-
2015
- 2015-01-16 CN CN201510024371.8A patent/CN105844608A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101101668A (en) * | 2007-07-13 | 2008-01-09 | 杭州电子科技大学 | Medical image cutting method based on oscillatory network |
CN101609504A (en) * | 2009-07-21 | 2009-12-23 | 华中科技大学 | A kind of method for detecting, distinguishing and locating infrared imagery sea-surface target |
CN102289668A (en) * | 2011-09-07 | 2011-12-21 | 谭洪舟 | Binaryzation processing method of self-adaption word image based on pixel neighborhood feature |
CN103310211A (en) * | 2013-04-26 | 2013-09-18 | 四川大学 | Filling mark recognition method based on image processing |
CN103903275A (en) * | 2014-04-23 | 2014-07-02 | 贵州大学 | Method for improving image segmentation effects by using wavelet fusion algorithm |
Non-Patent Citations (3)
Title |
---|
SHI ZHANG ET AL: ""Urinary Sediment Images Segmentation Based on Efficient Gabor flters"", 《2007 IEEE/ICME INTERNATIONAL CONFERENCE ON COMPLEX MEDICAL ENGINEERIN2\G》 * |
ZENG XIAOPING ET AL: ""Urinary Sediment Image Segmentation Based on Wavelet and Mathematical Morphology"", 《IMACS MULTICONFERENCE ON "COMPUTATIONAL ENGINEERING IN SYSTEMS APPLICATIONS"(CESA)》 * |
苏传朋: ""尿沉渣显微图像有形成分自动分割算法研究"", 《中国优秀博硕士学位论文全文数据库 (硕士) 医药卫生科技辑》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110717465A (en) * | 2019-10-15 | 2020-01-21 | 周兴祥 | Identification method and system based on calcium oxalate crystallization |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9704017B2 (en) | Image processing device, program, image processing method, computer-readable medium, and image processing system | |
JP5413408B2 (en) | Image processing apparatus, program, and image processing system | |
CN105512683B (en) | Object localization method and device based on convolutional neural networks | |
EP2523165B1 (en) | Image processing method and image processing device | |
JP6194791B2 (en) | Image processing apparatus and program | |
NL2024774B1 (en) | Blood leukocyte segmentation method based on adaptive histogram thresholding and contour detection | |
CN103093212A (en) | Method and device for clipping facial images based on face detection and face tracking | |
EP2889618A1 (en) | Image processing device, program, image processing method, computer-readable medium, and image processing system | |
US8548247B2 (en) | Image processing apparatus and method, and program | |
US10430943B2 (en) | Automated nuclei area/number estimation for IHC image analysis | |
CN103475800B (en) | Method and device for detecting foreground in image sequence | |
CN101984453B (en) | Human eye recognition system and method | |
US11908183B2 (en) | Image analysis and processing pipeline with real-time feedback and autocapture capabilities, and visualization and configuration system | |
US20130163844A1 (en) | Image processing apparatus, image processing method, non-transitory computer-readable medium, and image processing system | |
CN111369523B (en) | Method, system, equipment and medium for detecting cell stack in microscopic image | |
Singh et al. | Learning to Predict Video Saliency using Temporal Superpixels. | |
CN108009558B (en) | Target detection method and device based on multiple models | |
JP2011165170A (en) | Object detection device and program | |
CN105844608A (en) | Urinary sediment image segmentation method and urinary sediment image segmentation device | |
CN108985350B (en) | Method and device for recognizing blurred image based on gradient amplitude sparse characteristic information, computing equipment and storage medium | |
US20220383629A1 (en) | Label-free cell classification and screening system based on hybrid transfer learning | |
US20220114819A1 (en) | Cell tracking method, image processing device, and program | |
US10146042B2 (en) | Image processing apparatus, storage medium, and image processing method | |
CN107767401B (en) | Infrared target real-time tracking method and device based on nuclear correlation filtering | |
CN109034174B (en) | Cascade classifier training method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160810 |