CN102194203A - Method and equipment for reducing storage capacity in human face detection - Google Patents
Method and equipment for reducing storage capacity in human face detection Download PDFInfo
- Publication number
- CN102194203A CN102194203A CN2010101259640A CN201010125964A CN102194203A CN 102194203 A CN102194203 A CN 102194203A CN 2010101259640 A CN2010101259640 A CN 2010101259640A CN 201010125964 A CN201010125964 A CN 201010125964A CN 102194203 A CN102194203 A CN 102194203A
- Authority
- CN
- China
- Prior art keywords
- pixel
- image
- bit wide
- error
- pixel bit
- 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
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000001514 detection method Methods 0.000 title claims abstract description 18
- 238000009792 diffusion process Methods 0.000 claims abstract description 37
- 238000010586 diagram Methods 0.000 description 5
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 241001270131 Agaricus moelleri Species 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention relates to method and equipment for reducing the storage capacity in human face detection. The method comprises the following steps of: reducing pixel bit width of an image to be detected; diffusing the error of a gray value of each pixel point, which is generated in the process of reducing the pixel bit width, to adjacent pixel points by adopting an error diffusion method; solving an integral image and an integrated square image of the image subjected to the pixel bit width reduction and the error diffusion, wherein the storage capacity required by the integral image is (the pixel bit width plus logarithm2(W*H))*W*H bit; the storage capacity required by the integrated square image is (the pixel bit width*2+logarithm2(W*H))*W)*H bit; W is the pixel quantity of the image width; and H is the pixel quantity of image height. According to the method and equipment, the pixel bit width of the image is reduced and the error diffusion is carried out; and on the premise of no influence on a human face detection algorithm, the storage spaces required by the integral image and the integrated square image in a hardware chip are reduced and the cost of the chip is reduced.
Description
Technical field
The present invention relates to a kind of method and apparatus of the people's of reduction face detection of stored amount.
Background technology
Along with the popularization of video monitoring, human face detection tech and become more and more important.In various human face detection techs, adopt the human face detection tech accuracy rate of AdaBoost self-adaptive enhancement algorithm higher, and realized by hardware chip.This technology generally adopts the gray level image of 8 bit bit wides of CCD or cmos camera output as input.
Fig. 1 is based on the exemplary plot of the little feature that is adopted in the human face detection tech of AdaBoost self-adaptive enhancement algorithm.
As shown in Figure 1, in people's face detection processing procedure of this technology, this technology has adopted a lot of typical little features.
For example, a kind of template of leftmost little character representation.In processing, need obtain earlier white pixel in the corresponding region and, obtain again black patch the corresponding region in pixel with, then both are subtracted each other.On the same position of people's face and non-face picture, the size of the value that calculates is different, thus this slightly feature can be used for distinguishing people's face and non-face.This shows, in people's face detects, a large amount of computings relate to pixel of asking in certain zone and.In order to quicken this computation process, need a link that is specifically designed to calculated product partial image and square integral image, and the memory space that this link takies is many.
Fig. 2 is the synoptic diagram of calculated product partial image and square integral image.
As shown in Figure 2, image is divided into A, B, four zones of C, D, and the point in each regional lower right corner is respectively 1,2,3,4 points, the coordinate in the image lower right corner be (x, y).
Coordinate points (x, y), the value of integral image be by shown in the summation of the gray-scale value of being had a few in the rectangular area that constitutes between the image upper left corner and the lower right corner.For example, the gray-scale value summation of each point in the A of 1 value representation zone is noted by abridging and is I1; The gray-scale value summation of each point is designated as I2 in 2 value representation zone A+B; Similarly, be designated as I3 3 value; Value at 4 is designated as I4.So the gray-scale value summation of rectangular area D can be expressed as I1+I4-I2-I3.
Similarly, the integrated square image is in the summation of the each point gray-scale value of 1 value representation zone A square, note by abridging to be A, and be A+B 2 values, be A+C 3 values, be A+B+C+D 4 values.So the summation of the gray-scale value of rectangular area D square also can simply be drawn by the 1st, 2,3,4 integrated square image value.
Should be appreciated that an effable different colours number of pixel depends on the every pixel of bit (bpp, bitper pixel), represent each pixel to represent with 8 bits such as 8bpp, promptly the pixel bit wide is 8 bits.In general, the pixel bit wide can be referred to as the image bit wide.
If piece image is wide is the W pixel, and height is the H pixel, and the pixel bit wide is 8 bits; Then in integral image, the maximum bit wide that certain pixel may need is 8+log
2(W
*H), the memory space of therefore whole integral image needs is about (8+log
2(W
*H))
*W
*The H bit; In like manner, in the integrated square image, the maximum bit wide that certain pixel may need is 16+log
2(W
*H), the memory space of therefore whole integrated square image needs is about (16+log
2(W
*H))
*W
*The H bit.
From above-mentioned formula as can be seen, memory space that integral image and square integral image are required and pixel bit wide have very direct relation, the too high meeting of pixel bit wide causes integral image and the needed memory space of square integral image very big, thereby increases the cost of hardware chip.
Summary of the invention
The invention provides a kind of method and apparatus of the reduction people face detection of stored amount that can overcome the above problems.
In first aspect, a kind of method of the people's of reduction face detection of stored amount is provided, comprising: the pixel bit wide that reduces image to be detected; Adopt error-diffusion method produced each gray values of pixel points in reducing pixel bit wide process error diffusion to neighbor pixel; To having reduced the pixel bit wide and having carried out the image of error diffusion, ask for its integral image and square integral image; Wherein, the integral image required storage is (pixel bit wide+log
2(W
*H))
*W
*The H bit, integrated square image required storage is (pixel a bit wide
*2+log
2(W
*H))
*W
*H bit, W are the pixel quantity of picture traverse, and H is the pixel quantity of picture altitude.
Preferably, error-diffusion method be error diffusion that gray values of pixel points is produced in reducing the pixel bit wide to right-hand, the lower left of described pixel, below and on bottom-right four neighbor pixels.
Preferably, the error distribution ratio that is adopted in the error diffusion is 3: 3: 5: 5, and respectively right-hand, the lower left of corresponding described pixel, below and bottom-right neighbor pixel.
Preferably, the reduction of pixel bit wide is that the pixel bit wide is reduced to 6 bits from 8 bits.
In second aspect, a kind of equipment of the people's of reduction face detection of stored amount is provided, comprising: the module that reduces the pixel bit wide of image to be detected; Adopt error-diffusion method produced each gray values of pixel points in reducing pixel bit wide process the error diffusion module to the neighbor pixel; To having reduced the pixel bit wide and having carried out the image of error diffusion, ask for the module of its integral image and square integral image; Wherein, the integral image required storage is (pixel bit wide+log
2(W
*H))
*W
*The H bit, integrated square image required storage is (pixel a bit wide
*2+log
2(W
*H))
*W
*H bit, W are the pixel quantity of picture traverse, and H is the pixel quantity of picture altitude.
The present invention is by reducing the pixel bit wide of image to be checked in people's face testing process, saved storage space required when calculating integral image and square integral image in the hardware chip, reduced chip cost; And the error that produces in pixel bit wide reduction process is spread, reduced influence, thereby the energy of having kept image is constant substantially the result of calculation of integral image and square integral image.
Description of drawings
Below with reference to accompanying drawings specific embodiments of the present invention is described in detail, in the accompanying drawings:
Fig. 3 is the method synoptic diagram that reduces people's face detection hardware memory space according to an embodiment of the invention; And
Fig. 4 is the synoptic diagram of error-diffusion method according to an embodiment of the invention.
Embodiment
In the present invention, reduce calculated product partial image and the needed memory space of square integral image in the processing, when carrying out described calculating, adopt the lower image of precision in order to detect at people's face.For example, every pixel in the image is represented with 6 bits or 5 bits, reduced the precision of images.
Fig. 3 is the method synoptic diagram that reduces people's face detection hardware memory space according to an embodiment of the invention.
As shown in Figure 3, at first reduce the pixel bit wide of image to be detected; Then, the error diffusion that adopts error-diffusion method just to produce in the pixel bit wide is opened; At last, to having reduced the pixel bit wide and having carried out the image of error diffusion, ask for its integral image and square integral image.
Should be pointed out that the pixel bit wide that can adopt the whole bag of tricks to reduce image.Such as, gray values of pixel points can be converted to expectation pixel bit wide corresponding gray scale value, thereby reduce the pixel bit wide.
Fig. 4 is the synoptic diagram of error-diffusion method according to an embodiment of the invention.
For example, the image of 8 bit bit wides promptly has a pixel on 256 grades of gray level images, and its gray-scale value is 110 (0～255).If convert this pixel to 6 bits, the i.e. pixel value of 64 grades of gray scales, a kind of method is divided by 4 with this pixel value.So, the gray-scale value after the conversion is 110/4=27.5, is exactly 27 after the reservation integer-bit, visible 0.5 the error that exists after conversion.
In order to make image after the conversion, need the error diffusion of introducing in the transfer process be opened with error-diffusion method near original image.So-called error diffusion is exactly when pixel depth reduces, and the variation error of pixel color is spread apart.By error diffusion, make naked eyes when observing picture, the error of the set integral body of neighbor pixel diminishes, and more presses close to original image.
As shown in Figure 4, when the gray-scale value of previous 8 bit pixel points is P, become that its gray-scale value is P ' behind 6 bits, then the error d that introduces in this process (with respect to 8 bit grayscale value) is:
D=P-4
*P ' (multiply by 4 is because P ' is 6 bit values)
This error is multiplied by after the corresponding weights, and on 4 the adjacent pixels that are added to: right-hand, lower left, below, lower right, its corresponding weights is 3/16,3/16,5/16,5/16.Weight has been represented the error diffusion scheme that adopted, and promptly the error distribution ratio at the neighbor pixel of correspondence is 3: 3: 5: 5.
Should be appreciated that and error diffusion still less and/or on the more pixel, can also can be adopted other error distribution ratio to adjacent in error diffusion, such as 7: 3: 5: 1 and so on error distribution ratio.
For example, originally the gray-scale value of Xia Fang 8 bit pixel points is Q, and after the overlay error, its gray-scale value Q ' becomes:
Q’＝Q+d
＊5/16
The rest may be inferred, can draw the gray-scale value of other neighbor pixels.For example, after overlay error, this pixel is right-hand, the gray-scale value of lower left and bottom-right neighbor pixel is respectively: Q+d
*3/16, Q+d
*3/16, Q+d
*5/16.By such error diffusion, and then can obtain a fabric width, high identical with original image, the pixel bit wide is the image of 6 bits.Then, to the pixel bit wide be its integral image of image calculation and square integral image of 6 bits.From last calculating to integral image and square integral image as can be known, before reducing the pixel bit wide, the memory space of integral image is (8+log
2(W
*H))
*W
*H bit, the memory space of integrated square image are (16+log
2(W
*H))
*W
*H, after the pixel bit wide that reduces image, the memory space of integral image is reduced to (6+log
2(W
*H))
*W
*The H bit, the memory space of integrated square image is reduced to (12+log
2(W
*H))
*W
*The H bit.This shows that integral image and square integral image required storage have had tangible reduction after the pixel bit wide reduces.
By adopting error-diffusion method, make the pixel bit wide of image reduce the integral image influence less.Because in certain zone, the error that each pixel brings because reduce bit wide generally all is diffused on the interior adjacent pixels point of the same area.In addition, with respect to concerning the influence of integral image, adopt error-diffusion method big slightly, but generally also can ignore the influence of integrated square image.Adopt error-diffusion method to lower the pixel bit wide of image, can keep the energy of image constant substantially.
Like this, do not influencing under the prerequisite of people's face detection algorithm substantially, saving hardware chip and preserved integral image and the needed storage space of square integral image, reducing the cost of chip.
In addition, the pixel bit wide of image can also be made simplification to corresponding arithmetic element (as multiplier, divider etc.) after reducing.
In other examples, also the image of 8 bit bit wides can be reduced to the image of 5 bits, 4 bits or other bit wides.Should be appreciated that according to concrete application scenario, can choose suitable bit wide.
Obviously, under the prerequisite that does not depart from true spirit of the present invention and scope, the present invention described here can have many variations.Therefore, the change that all it will be apparent to those skilled in the art that all should be included within the scope that these claims contain.The present invention's scope required for protection is only limited by described claims.
Claims (5)
1. method that reduces people's face detection of stored amount comprises:
Reduce the pixel bit wide of image to be detected;
Adopt error-diffusion method produced each gray values of pixel points in reducing pixel bit wide process error diffusion to neighbor pixel;
To having reduced the pixel bit wide and having carried out the image of error diffusion, ask for its integral image and square integral image;
Wherein, the integral image required storage is (pixel bit wide+log
2(W*H)) * W*H bit, integrated square image required storage are (pixel bit wide * 2+log
2(W*H)) * W*H bit, W are the pixel quantity of picture traverse, and H is the pixel quantity of picture altitude.
2. method according to claim 1, wherein, error-diffusion method be error diffusion that gray values of pixel points is produced in reducing the pixel bit wide to right-hand, the lower left of described pixel, below and on bottom-right four neighbor pixels.
3. method according to claim 2, wherein, the error distribution ratio that is adopted in the error diffusion is 3: 3: 5: 5, respectively right-hand, the lower left of corresponding described pixel, below and bottom-right neighbor pixel.
4. method according to claim 1, wherein, it is that the pixel bit wide is reduced to 6 bits from 8 bits that the pixel bit wide reduces.
5. equipment that reduces people's face detection of stored amount comprises:
Reduce the module of the pixel bit wide of image to be detected;
Adopt error-diffusion method produced each gray values of pixel points in reducing pixel bit wide process the error diffusion module to the neighbor pixel;
To having reduced the pixel bit wide and having carried out the image of error diffusion, ask for the module of its integral image and square integral image;
Wherein, the integral image required storage is (pixel bit wide+log
2(W*H)) * W*H bit, integrated square image required storage are (pixel bit wide * 2+log
2(W*H)) * W*H bit, W are the pixel quantity of picture traverse, and H is the pixel quantity of picture altitude.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010125964.0A CN102194203B (en) | 2010-03-16 | 2010-03-16 | A kind of method and apparatus reducing Face datection memory space |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010125964.0A CN102194203B (en) | 2010-03-16 | 2010-03-16 | A kind of method and apparatus reducing Face datection memory space |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102194203A true CN102194203A (en) | 2011-09-21 |
| CN102194203B CN102194203B (en) | 2015-09-02 |
Family
ID=44602230
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010125964.0A Active CN102194203B (en) | 2010-03-16 | 2010-03-16 | A kind of method and apparatus reducing Face datection memory space |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102194203B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107154233A (en) * | 2017-04-19 | 2017-09-12 | 西安诺瓦电子科技有限公司 | Image processing method and device and display control card |
| CN107644207A (en) * | 2016-06-27 | 2018-01-30 | 广东欧珀移动通信有限公司 | A kind of fingerprint image processing method and Related product |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101178770A (en) * | 2007-12-11 | 2008-05-14 | 北京中星微电子有限公司 | Image detection method and apparatus |
-
2010
- 2010-03-16 CN CN201010125964.0A patent/CN102194203B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101178770A (en) * | 2007-12-11 | 2008-05-14 | 北京中星微电子有限公司 | Image detection method and apparatus |
Non-Patent Citations (1)
| Title |
|---|
| 林志贤等: "场致发射显示器新型图像驱动技术的研究", 《液晶与显示》, vol. 24, no. 5, 31 October 2009 (2009-10-31), pages 685 - 691 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107644207A (en) * | 2016-06-27 | 2018-01-30 | 广东欧珀移动通信有限公司 | A kind of fingerprint image processing method and Related product |
| CN107644207B (en) * | 2016-06-27 | 2021-03-12 | Oppo广东移动通信有限公司 | Fingerprint image processing method and related product |
| CN107154233A (en) * | 2017-04-19 | 2017-09-12 | 西安诺瓦电子科技有限公司 | Image processing method and device and display control card |
| CN107154233B (en) * | 2017-04-19 | 2020-07-14 | 西安诺瓦星云科技股份有限公司 | Image processing method and device and display control card |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102194203B (en) | 2015-09-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104992657B (en) | Mura compensating modules and method, display device and method | |
| US8675970B2 (en) | Image processing apparatus, image processing method, and image processing program | |
| CN101901480B (en) | Device and method for processing image data | |
| US8131111B2 (en) | Device of processing dead pixel | |
| CN107680056A (en) | A kind of image processing method and device | |
| CN106570909B (en) | Skin color detection method, device and terminal | |
| CN104182927B (en) | Picture watermark adding method and device | |
| EP1789922A2 (en) | Apparatus and method for processing images | |
| JP2013042415A (en) | Image processing apparatus, image processing method, and computer program | |
| CN111340721B (en) | Pixel correction method, device, equipment and readable storage medium | |
| WO2019041842A1 (en) | Image processing method and device, storage medium and computer device | |
| KR20150120975A (en) | Unevenness inspection system, unevenness inspection method, and unevenness inspection program | |
| CN108074237B (en) | Image definition detection method and device, storage medium and electronic equipment | |
| CN108109147A (en) | A kind of reference-free quality evaluation method of blurred picture | |
| CN100493134C (en) | Method and system for processing image | |
| CN104915939A (en) | Image equalization method and device | |
| US20090274368A1 (en) | Image correction method and apparatus | |
| CN108305232A (en) | A kind of single frames high dynamic range images generation method | |
| KR101715489B1 (en) | Image generating device and image generating method | |
| CN106651816B (en) | Method and system for halftone processing of digital image | |
| CN108805873A (en) | Image processing method and device | |
| CN102194203A (en) | Method and equipment for reducing storage capacity in human face detection | |
| US20130259373A1 (en) | Image processing device capable of determining types of images accurately | |
| CN116823674B (en) | Cross-modal fusion underwater image enhancement method | |
| CN108073964B (en) | Perspective two-dimensional code generation 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 | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20180105 Address after: 100083 Haidian District, Xueyuan Road, No. 35, the world building, the second floor of the building on the ground floor, No. 16 Patentee after: Zhongxing Technology Co., Ltd. Address before: 100083 Haidian District, Xueyuan Road, No. 35, the world building, the second floor of the building on the ground floor, No. 16 Patentee before: Beijing Vimicro Corporation Effective date of registration: 20180105 Address after: 100083 Haidian District, Xueyuan Road, No. 35, the world building, the second floor of the building on the ground floor, No. 16 Patentee after: Zhongxing Technology Co., Ltd. Address before: 100083 Haidian District, Xueyuan Road, No. 35, the world building, the second floor of the building on the ground floor, No. 16 Patentee before: Beijing Vimicro Corporation |
|
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 100083 Haidian District, Xueyuan Road, No. 35, the world building, the second floor of the building on the ground floor, No. 16 Patentee after: Mid Star Technology Limited by Share Ltd Address before: 100083 Haidian District, Xueyuan Road, No. 35, the world building, the second floor of the building on the ground floor, No. 16 Patentee before: Zhongxing Technology Co., Ltd. Address after: 100083 Haidian District, Xueyuan Road, No. 35, the world building, the second floor of the building on the ground floor, No. 16 Patentee after: Mid Star Technology Limited by Share Ltd Address before: 100083 Haidian District, Xueyuan Road, No. 35, the world building, the second floor of the building on the ground floor, No. 16 Patentee before: Zhongxing Technology Co., Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200514 Address after: No. 2201-2206, 22 / F, block a, Shanghai ASEAN business building, No. 779, Chunrong street, Chenggong District, Kunming City, Yunnan Province 650000 Patentee after: Yunnan Zhongxing Electronic Co., Ltd Address before: 100083, Haidian District, Xueyuan Road, Beijing No. 35, Nanjing Ning building, 16 floor Patentee before: Mid Star Technology Limited by Share Ltd. |