WO2015082957A1 - Adaptive integral histogram calculation for image thresholding - Google Patents
Adaptive integral histogram calculation for image thresholding Download PDFInfo
- Publication number
- WO2015082957A1 WO2015082957A1 PCT/IB2013/060587 IB2013060587W WO2015082957A1 WO 2015082957 A1 WO2015082957 A1 WO 2015082957A1 IB 2013060587 W IB2013060587 W IB 2013060587W WO 2015082957 A1 WO2015082957 A1 WO 2015082957A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- integral histogram
- equals
- calculation
- sorting
- histogram calculation
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration by the use of histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- 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
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
Definitions
- histogram of the image is calculated by counting the number of the gray levels while scanning the image. The total number of the each gray level is p(i) stored in histogram buffer (FIG. l).
- integral histogram is calculated by reading the histogram values.
- the entries q(i) of the integral histogram buffer holds the total number of pixels which are in the range 0 to i, inclusively (EQ. l). EQ. l
- histogram buffer should be read entry by entry. It means that 2 N read operations are required where N is the total bit number to define the gray level.
- Table 1 shows the necessary read cycles and the corresponding time to calculate integral histogram. (For time calculations read clock equals to 100MHZ - 10ns)
- the video frame (image) duration is around 20ms.
- integral histogram calculation time values are small and no effect on real-time flow.
- FIG.3 is the thresholding operation.
- FIG.7 is the flowchart of the sub-step of setting the relation buffer.
- FIG.8 is the flowchart of the step of extracting different gray levels.
- FIG.9 shows the step of calculating the integral histogram.
- FIG.10 is the flowchart of the sub-step of integral histogram calculation with sorting.
- FIG.ll is the flowchart of the sub-step of sorting the different gray values.
- FIG.12 is the flowchart of the sub-step of integral histogram calculation with jumping.
- a method for adaptive integral histogram calculation (100) comprises the steps of; receiving an image or a patch (101),
- the relation buffer (FIG.5) stores the relations between the gray levels.
- Relation buffer contains T columns (nx) and 2 N -1 rows (gl-x).
- the rows gl-x represents the possible gray levels.
- T is the depth of the relation and it is a predetermined value.
- gl-0 I nl store the relation between gl-0 and gl-1.
- the gl-0/nl is set to logic "one”. Namely, the buffer stores which gray level is valid next.
- the step "constructing the neighborhood relation (104)" comprises the sub-steps of;
- the sub-step "setting the relation buffer (203)" comprises the sub-steps of
- adaptive integral histogram calculation (100) method checks the number of different gray levels and calculates the integral histogram (106) in two ways; sorting (108) or jumping (107). If the number of different gray values is less than a pre-determined value, integral histogram is calculated with sorting (108). The number of different gray levels is calculated while extracting different gray levels (105)
- the step "extracting different gray levels (105),” comprises the sub-steps of;
- the step "integral histogram calculation with sorting (108)" comprises the sub-steps of
- the sub-step “sorting the different gray values (502),” comprises the sub-steps of
- integral histogram calculation with sorting (108) starts the calculation from the minimum gray value and continues the calculation with the sorted gray values (503) until the maximum gray value.
- Table 3 shows the required cycles to complete sorting and Table 4 shows the integral histogram calculation time for different number of gray values when sorting is used.
- integral histogram is calculated with jumping (107).
- the step "integral histogram calculation with jumping (107)" comprises the sub-steps of
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
- Facsimile Image Signal Circuits (AREA)
Abstract
Image thresholding is very crucial for accurate detection of the target. Most of the good thresholding techniques require integral histogram calculations at the patch level instead of frame level. This invention shortens the integral histogram calculation time to process many small patches while fulfilling the real-time constraints.
Description
DESCRIPTION
ADAPTIVE INTEGRAL HISTOGRAM CALCULATION FOR IMAGE THRESHOLDING
Field of the invention
This invention is related with a method to shorten integral histogram calculation time.
Background of the invention
Image thresholding is used for many image processing applications such as image segmentation and target detection. There are many different image thresholding techniques in use. These techniques have different accuracy performance and complexity. Most of the powerful thresholding algorithms require recursive calculation of a cost function. Besides this, cost functions commonly need integral histogram values.
To calculate the integral histogram, firstly histogram of the image should be calculated. Histogram of the image is calculated by counting the number of the gray levels while scanning the image. The total number of the each gray level is p(i) stored in histogram buffer (FIG. l).
(p(i) : holds the number of the gray level in the scanned image. For example p(0) holds the number of "0" value pixels in the image.)
After scanning the entire image and finding the histogram of the image, integral histogram is calculated by reading the histogram values. The entries q(i) of the integral histogram buffer holds the total number of pixels which are in the range 0 to i, inclusively (EQ. l).
EQ. l
To calculate the integral histogram, histogram buffer should be read entry by entry. It means that 2N read operations are required where N is the total bit number to define the gray level.
(N: total bit number to define the gray level. For example for 256 gray level, N = 8.)
Table 1 shows the necessary read cycles and the corresponding time to calculate integral histogram. (For time calculations read clock equals to 100MHZ - 10ns)
For a 640x512 @ 50Hz video sequence, the video frame (image) duration is around 20ms. When compared with the total frame duration, integral histogram calculation time values are small and no effect on real-time flow.
But for some applications such as thermal target detection, instead of a whole image, integral histogram is calculated for small patches. Patch is a part of the image which has a lower resolution. These patches can have different dimensions. (FIG.2)
For accurate detection of the target, image thresholding is very crucial. Because there can be noise around actual target and this noise can prevent the accurate detection. Good thresholding can remove the noise and can generate a good binary patch which is suitable for the detection (FIG.3). But most of the good thresholding techniques require integral histogram calculations at the patch level instead of image level.
In an image, there can be many targets and as a result many patches. For the different patch dimensions Table 2 shows how many patches can be realized in a 640x512 image and the resultant integral image calculation time for clock equals to 100MHZ.
Table 2
From Table 2, it is obvious that standard integral histogram calculation is not suitable for real time implementations. The invention proposes new approaches to reduce these calculation time values.
Objects of the invention
This invention enables to calculate integral histogram in a shorter time. Detailed description of the invention
Adaptive integral histogram calculation for image thresholding method in order to fulfill the objects of the present invention is illustrated in the attached figures, where:
FIG.l shows the histogram and integral histogram calculation flow.
FIG.2 shows the targets and patches on the image.
FIG.3 is the thresholding operation.
FIG.4 is the flowchart of the method for adaptive integral histogram calculation. FIG.5 shows the relation buffer.
FIG.6 is the flowchart of the step of constructing the neighborhood relation.
FIG.7 is the flowchart of the sub-step of setting the relation buffer.
FIG.8 is the flowchart of the step of extracting different gray levels.
FIG.9 shows the step of calculating the integral histogram.
FIG.10 is the flowchart of the sub-step of integral histogram calculation with sorting.
FIG.ll is the flowchart of the sub-step of sorting the different gray values.
FIG.12 is the flowchart of the sub-step of integral histogram calculation with jumping.
A method for adaptive integral histogram calculation (100) comprises the steps of; receiving an image or a patch (101),
calculating the histogram (102),
calculating the minimum and maximum gray level (103),
constructing the neighborhood relation (104),
extracting different gray levels (105),
calculating the integral histogram (106),
In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method can be used for either an image or a patch. After receiving the image or the patch (101) the captured gray values (GV) are used by the steps of (102), (103), (104) and (105).
The histogram buffer (FIG. l) is obtained by calculating the histogram (102) of the image or patch. The minimum and maximum gray levels are calculated (103) by comparing the captured gray values.
The relation buffer (FIG.5) stores the relations between the gray levels. Relation buffer contains T columns (nx) and 2N-1 rows (gl-x). The rows gl-x represents the possible gray levels. T is the depth of the relation and it is a predetermined value. For example gl-0 I nl store the relation between gl-0 and gl-1. In the calculated image or patch, if there exists a gray level equals to "1", the gl-0/nl is set to logic "one". Namely, the buffer stores which gray level is valid next.
In the method for adaptive integral histogram calculation (100), the step "constructing the neighborhood relation (104)" comprises the sub-steps of;
resetting the relation buffer to logic "zero" (201),
reading the captured gray value (GV) (202),
setting the relation buffer (203),
checking the end of the image or patch (204),
returning (202) if the read gray value does not belong to the last pixel in the image or patch (205),
finishing the construction if the read gray value belongs to last pixel in the image or patch (206)
In the preferred embodiment of the invention, the sub-step "setting the relation buffer (203)" comprises the sub-steps of
assigning the value of x equals to "1"
setting gl-( GV-x) I nx to logic "one"
- adding "1" to x
checking the value of x
returning (302) if x does not equal to "T+l"
finishing the set operation if x equals to "T+l"
In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method checks the number of different gray levels and calculates the integral histogram (106) in two ways; sorting (108) or jumping (107). If the number of different gray values is less than a pre-determined value, integral histogram is calculated with sorting (108). The number of different gray levels is calculated while extracting different gray levels (105)
In the method for adaptive integral histogram calculation (100), the step "extracting different gray levels (105)," comprises the sub-steps of;
setting the different gray level number to "0" (401),
reading the captured gray value (GV) (202),
reading the related entry p(GV) of the histogram buffer (402),
- checking the value of p(GV) (403),
- adding "1" to different gray level number if p(GV) equals to "0" (404), storing the captured gray value in the sorting buffer if p(GV) equals to "0" (405),
applying no change on the different gray level number and sorting buffer if p(GV) does not equal to "0" (406),
checking the different gray level number (407),
finishing the calculation if different gray level number exceeds predetermined value (408),
checking the end of the image or patch (204),
returning (202) if the read gray value does not belong to the last pixel in the image or patch (205),
finishing the calculation if the read gray value belongs to last pixel in the image or patch (409),
In the preferred embodiment of the invention, the step "integral histogram calculation with sorting (108)," comprises the sub-steps of
reading the different gray values from the sorting buffer (501),
sorting the different gray values (502),
calculating the integral histogram with the sorted gray values (503),
In the preferred embodiment of the invention, the sub-step "sorting the different gray values (502)," comprises the sub-steps of
forming n arrays of length m for an n x m block
dividing each array into 2 equal parts {mil)
sorting each divided array
merging divided arrays while sorting
merging each array while sorting
In the preferred embodiment of the invention, integral histogram calculation with sorting (108) starts the calculation from the minimum gray value and continues the calculation with the sorted gray values (503) until the maximum gray value.
Table 3 shows the required cycles to complete sorting and Table 4 shows the integral histogram calculation time for different number of gray values when sorting is used.
Table 3
Table 4
64 75 64 0.00141
100 113 100 0.00216
144 159 144 0.00307
256 275 256 0.00537
If the number of different gray values is higher than a pre-determined value, integral histogram is calculated with jumping (107).
In the preferred embodiment of the invention, the step "integral histogram calculation with jumping (107)," comprises the sub-steps of
reading the minimum gray value (MGV) (701),
setting i equals to MGV and i_valid equals to MGV (702),
setting q(z) equals to p(z) (703),
reading gl-i from relation buffer (704),
assigning the value of x equals to "1" (301),
jumping to gray value i + x if gl-i I nx equals logic "one" (705), setting q(z+x) equals q(i_valid) plus p(z+x) and i equals to i+x and i_valid equals to i+x (706),
returning to (704) if i does not equal to maximum gray value else moving to (711) (707),
adding "1" to x if gl-i I nx equals logic "zero" (708),
- returning to (705) if x does not equals to "T+l" (709),
setting i equals to i+x and returning (704) (710),
finishing the calculation (711),
Table 5 shows the possible integral histogram calculation times for the different T values for N=14 and clock=100MHZ when jumping is used.
Table.5
32 512 0.00512
64 256 0.00256
Table 6 shows how many patches can be realized in a 640x512 image and the resultant integral image calculation time for clock equals to 100MHZ and T =32.
Table.6
Table 6 shows that there is a considerable reduction in the calculation times between the standard methods and the proposed method. Adaptive structure selects between the methods (sort or jump) according to the number of different gray levels
Claims
1. A method for adaptive integral histogram calculation (100) comprises the steps of;
receiving an image or a patch (101),
calculating the histogram (102),
calculating the minimum and maximum gray level (103),
constructing the neighborhood relation (104),
extracting different gray levels (105),
calculating the integral histogram (106),
2. In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method according to claim 1 characterized by using the captured gray values (GV) which are received in an image or a patch (101), at the steps of (102), (103), (104) and (105).
3. In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method according to claim 1 characterized by generating the histogram buffer by calculating the histogram (102) from the captured gray values.
4. In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method according to claim 1 characterized by calculating the minimum and maximum gray value (103) by comparing the captured gray values.
5. In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method according to claim 1 characterized by using the minimum and maximum gray values at the step of calculating the integral histogram (106),
6. The adaptive integral histogram calculation (100) method according to claim 1 characterized by generating a relation buffer which contains T columns (nx) and 2N-1 rows (gl-x) by constructing a neighborhood relation (104).
7. In the method for adaptive integral histogram calculation (100), the step "constructing the neighborhood relation (104)" comprises the sub-steps of;
resetting the relation buffer to logic "zero" (201),
reading the captured gray value (GV) (202),
setting the relation buffer (203),
checking the end of the image or patch (204),
returning (202) if the read gray value does not belong to the last pixel in the image or patch (205),
finishing the construction if the read gray value belongs to last pixel in the image or patch (206)
8. In the preferred embodiment of the invention, the sub-step "setting the relation buffer (203)" comprises the sub-steps of;
assigning the value of x equals to "1"
setting gl-(GV-x) I nx to logic "one"
- adding "1" to x
checking the value of x
returning (302) if x does not equal to "T+l"
finishing the set operation if x equals to "T+l"
9. In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method according to claim 1 characterized by using the relation buffer at the sub-step of integral histogram calculation with jumping (107),
10. In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method according to claim 1 characterized by calculating the integral histogram (106) in two ways; sorting (108) or jumping (107).
11. In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method according to claim 10 characterized by selecting the integral histogram calculation type; sorting (108) or jumping (107) by evaluating the number of different gray values which is obtained at the step of extracting different gray levels (105).
12. In the method for adaptive integral histogram calculation (100), the step "extracting different gray levels (105)," comprises the sub-steps of; setting the different gray level number to "0" (401),
reading the captured gray value (GV) (202),
reading the related entry p(GV) of the histogram buffer (402), - checking the value of p(GV) (403),
adding "1" to different gray level number if p(GV) equals to "0" (404),
storing the captured gray value in the sorting buffer if p(GV) equals to "0" (405),
applying no change on the different gray level number and sorting buffer if p(GV) does not equal to "0" (406),
checking the different gray level number (407),
finishing the calculation if different gray level number exceeds predetermined value (408),
checking the end of the image or patch (204),
returning (202) if the read gray value does not belong to the last pixel in the image or patch (205),
finishing the calculation if the read gray value belongs to last pixel
in the image or patch (409),
13. In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method according to claim 10 characterized by calculating the integral histogram (106) with sorting (108), if the number of different gray values is less than a predetermined value.
14. In the preferred embodiment of the invention, the step "integral histogram calculation with sorting (108)," comprises the sub-steps of reading the different gray values from the sorting buffer (501), sorting the different gray values (502),
calculating the integral histogram with the sorted gray values (503),
15. In the preferred embodiment of the invention, the sub-step "sorting the different gray values (502)," comprises the sub-steps of
forming n arrays of length m for an n x m block
dividing each array into 2 equal parts {mil)
sorting each divided array
merging divided arrays while sorting
merging each array while sorting
16. In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method according to claim 10 characterized by calculating the integral histogram (106) with jumping (107), if the number of different gray values is higher than a predetermined value.
17. In the preferred embodiment of the invention, the step "integral histogram calculation with jumping (107)," comprises the sub-steps of
reading the minimum gray value (MGV) (701),
- setting i equals to MGV and i_valid equals to MGV (702),
setting q(z) equals to p(z') (703),
reading gl-i from relation buffer (704),
assigning the value of x equals to "1" (301),
- jumping to gray value i + x if gl-i / nx equals logic "one" (705), setting q( +x) equals q(i_valid) plus p( +x) and equals to i+x and i_valid equals to i+x (706),
returning to (704) if i does not equal to maximum gray value else moving to (711) (707),
adding "1" to x if gl-i I nx equals logic "zero" (708),
- returning to (705) if x does not equals to "T+l" (709),
setting i equals to i+x and returning (704) (710),
finishing the calculation (71 1), In the preferred embodiment of the invention, adaptive integral histogram calculation (100) method according to claim 5 characterized by starting the integral histogram calculation from the minimum gray value and continuing the calculation until the maximum gray value.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/IB2013/060587 WO2015082957A1 (en) | 2013-12-03 | 2013-12-03 | Adaptive integral histogram calculation for image thresholding |
| KR1020157033252A KR101863999B1 (en) | 2013-12-03 | 2013-12-03 | Adaptive integral histogram calculation for image thresholding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/IB2013/060587 WO2015082957A1 (en) | 2013-12-03 | 2013-12-03 | Adaptive integral histogram calculation for image thresholding |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2015082957A1 true WO2015082957A1 (en) | 2015-06-11 |
Family
ID=49943421
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/IB2013/060587 WO2015082957A1 (en) | 2013-12-03 | 2013-12-03 | Adaptive integral histogram calculation for image thresholding |
Country Status (2)
| Country | Link |
|---|---|
| KR (1) | KR101863999B1 (en) |
| WO (1) | WO2015082957A1 (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006083023A1 (en) * | 2005-02-07 | 2006-08-10 | Mitsubishi Denki Kabushiki Kaisha | Computer implemented method for extracting integral histogram from sampled data |
| US20070133878A1 (en) * | 2005-12-14 | 2007-06-14 | Porikli Fatih M | Method for constructing covariance matrices from data features |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101349968B1 (en) * | 2011-11-28 | 2014-01-14 | 네이버 주식회사 | Image processing apparatus and method for automatically adjustment of image |
-
2013
- 2013-12-03 KR KR1020157033252A patent/KR101863999B1/en active IP Right Grant
- 2013-12-03 WO PCT/IB2013/060587 patent/WO2015082957A1/en active Application Filing
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006083023A1 (en) * | 2005-02-07 | 2006-08-10 | Mitsubishi Denki Kabushiki Kaisha | Computer implemented method for extracting integral histogram from sampled data |
| US20070133878A1 (en) * | 2005-12-14 | 2007-06-14 | Porikli Fatih M | Method for constructing covariance matrices from data features |
Non-Patent Citations (3)
| Title |
|---|
| BILGIC B ET AL: "Efficient integral image computation on the GPU", INTELLIGENT VEHICLES SYMPOSIUM (IV), 2010 IEEE, IEEE, PISCATAWAY, NJ, USA, 21 June 2010 (2010-06-21), pages 528 - 533, XP031827364, ISBN: 978-1-4244-7866-8 * |
| ONCEL TUZEL ET AL: "Region Covariance: A Fast Descriptor for Detection and Classification", 1 January 2006, COMPUTER VISION - ECCV 2006 LECTURE NOTES IN COMPUTER SCIENCE;;LNCS, SPRINGER, BERLIN, DE, PAGE(S) 589 - 600, ISBN: 978-3-540-33834-5, XP019036470 * |
| PORIKLI FATIH: "Integral Histogram: A Fast Way To Extract Histograms in Cartesian Spaces", PROCEEDINGS / 2005 IEEE COMPUTER SOCIETY CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2005 : [20 - 25 JUNE 2005, SAN DIEGO, CA], IEEE, PISCATAWAY, NJ, USA, vol. 1, 20 June 2005 (2005-06-20), pages 829 - 836, XP010817358, ISBN: 978-0-7695-2372-9, DOI: 10.1109/CVPR.2005.188 * |
Also Published As
| Publication number | Publication date |
|---|---|
| KR20160003016A (en) | 2016-01-08 |
| KR101863999B1 (en) | 2018-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110598558B (en) | Crowd density estimation method, device, electronic equipment and medium | |
| CN110991311B (en) | Target detection method based on dense connection deep network | |
| CN109859227B (en) | Method and device for detecting flip image, computer equipment and storage medium | |
| CN100371676C (en) | Method and device for quick high precision positioning light spot image mass center | |
| CN108154080B (en) | Method for quickly tracing to source of video equipment | |
| CN110049309B (en) | Method and device for detecting stability of image frame in video stream | |
| US8811750B2 (en) | Apparatus and method for extracting edge in image | |
| CN110321913B (en) | Text recognition method and device | |
| CN109492577A (en) | A kind of gesture identification method, device and electronic equipment | |
| Priya et al. | Edge strength extraction using orthogonal vectors for shot boundary detection | |
| CN110555863A (en) | moving object detection method and device and computer readable storage medium | |
| CN111967345B (en) | Method for judging shielding state of camera in real time | |
| CN104394411A (en) | Median filtering device and method | |
| CN108305265B (en) | Real-time processing method and system for weak and small target image | |
| WO2015031350A1 (en) | Systems and methods for memory utilization for object detection | |
| CN111159150A (en) | Data expansion method and device | |
| CN104881670B (en) | A kind of fast target extracting method for SAR orientation angular estimation | |
| CN108197531B (en) | Road curve detection method, device and terminal | |
| CN111753729A (en) | False face detection method and device, electronic equipment and storage medium | |
| CN107958226B (en) | Road curve detection method, device and terminal | |
| WO2015082957A1 (en) | Adaptive integral histogram calculation for image thresholding | |
| CN112766256B (en) | Grating phase diagram processing method and device, electronic equipment and storage medium | |
| CN113313112B (en) | Image processing method and device, computer equipment and storage medium | |
| Hagara et al. | FPGA Implementation of Unimodal Thresholding | |
| CN109325945B (en) | Image processing method, image processing device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 13818802 Country of ref document: EP Kind code of ref document: A1 |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 2015/1244.1 Country of ref document: KZ |
|
| ENP | Entry into the national phase |
Ref document number: 20157033252 Country of ref document: KR Kind code of ref document: A |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 13818802 Country of ref document: EP Kind code of ref document: A1 |