FORM 2
THE PATENTS ACT, 1970
(39 of 1970)
&
The Patents Rules, 2003
COMPLETE SPECIFICATION
(See section 10 and rule 13)
1.
Title : METHOD AND SYSTEM FOR SEGMENTING HUMAN SKIN COLOR REGIONS
ON AN IMAGE
2. APPLICANT (S)
a)
NAME
Samsung Electronics Company
b)
NATIONALITY
Republic of Korea
c)
ADDRESS
Samsung Electronics Company
416 Maetan-Dong, Yeongtong-GU, SUWON-SI,
Gyeonggi-do 442-742,
Republic of Korea
3. PREAMBLE TO THE DESCRIPTION
PROVISIONAL
The following specification describes the invention
��
COMPLETE
The following specification particularly describes the invention and the manner in which it is to be performed.
4. DESCRIPTION (Description shall start from next page)
5. CLAIMS (not applicable for provisional specification. Claims should start with the preamble – “I/we claim” on separate page)
6. DATE AND SIGNATURE (to be given at the end of last page of specification)
7. ABSTRACT OF THE INVENTION (to be given on separate page)
1/20
METHOD AND SYSTEM FOR SEGMENTING HUMAN SKIN COLOR REGIONS ON AN IMAGE
FIELD
[0001]
The present disclosure relates generally to the field of visual image processing. More particularly, the present disclosure relates to a method and a system for segmenting human skin color regions on an image.
BACKGROUND
[0002]
In the current scenario, real time image processing applications are used to segment one or more regions of a visual image. The visual image is processed in real time to generate segmented regions with desired characteristics. For example, the segmentation of the visual image can be applied for locating objects, finger print recognition, face recognition, and for medical imaging. However, processing time required to process the visual image in real-time is directly proportional to complexity of image segmentation desired. Further, the real time image processing applications may be executed using fixed programs with a probability of incorrectly segmenting the visual image.
[0003]
In light of the foregoing discussion there is a need for a method and system for segmenting human skin color regions on an image by reducing the processing time and improving the accuracy of image segmentation.
SUMMARY
2/20
[0004]
Embodiments of the present disclosure described herein provide a method and system for segmenting human skin color regions on an image.
[0005]
An example of a method for segmenting human skin color regions on an image includes extracting fuzzy rules corresponding to an image data set based on processing parameters, wherein the processing parameters indicate human skin color regions and non-human skin color regions associated with the image data set. The method also includes converting the fuzzy rules to radial basis function rules. Further, the method includes applying the radial basis function rules on a visual image. Further, the method also includes rendering the human skin color regions on the visual image in a first distinctive color and the non- human skin color regions on the visual image in a second distinctive color based on the applying.
[0006]
An electronic device for segmenting human skin color regions on an image includes a communication interface that receives an input associated with at least one of the human skin color regions and the non-human skin color regions from one or more portions on a visual image. The electronic device also includes a memory for storing information. Further, the electronic device includes a processor responsive to the information to extract fuzzy rules corresponding to an image data set based on processing parameters, wherein the processing parameters indicate human skin color regions and non- human skin color regions associated with the image data set. The processor also converts the fuzzy rules to radial basis function rules. Further, the processor applies the radial basis function rules on a visual image. Further, the processor renders human skin color regions on the visual image in a first distinctive color and the non- human skin color regions in a second distinctive color. 3/20
BRIEF DESCRIPTION OF FIGURES
[0007]
The accompanying figures, similar reference numerals may refer to identical or functionally similar elements. These reference numerals are used in the detailed description to illustrate various embodiments and to explain various aspects and advantages of the present disclosure.
[0008]
FIG. 1 is a block diagram of an electronic device for segmenting human skin color regions on an image, in accordance with which various embodiments can be implemented;
[0009]
FIG. 2 (Prior Art) is an exemplary illustration of a fuzzy decision tree;
[0010]
FIG. 3 (Prior Art) is an exemplary illustration of generalized Gaussian radial basis function network.
[0011]
FIG. 4a - FIG. 4b is a flow chart illustrating a method for segmenting human skin color regions on an image, in accordance with one embodiment; and
[0012]
FIG. 5 is an exemplary illustration of segmenting human skin color regions on an image, in accordance with one embodiment.
[0013]
Persons skilled in the art will appreciate that elements in the figures are illustrated for simplicity and clarity and may have not been drawn to scale. For example, the dimensions of some of the elements in the figures may be exaggerated relative to other elements to help to improve understanding of various embodiments of the present disclosure.
4/20
DETAILED DESCRIPTION
[0014]
It should be observed that method steps and system components have been represented by conventional symbols in the figures, showing only specific details that are relevant for an understanding of the present disclosure. Further, details that may be readily apparent to person ordinarily skilled in the art may not have been disclosed. In the present disclosure, relational terms such as first and second, and the like, may be used to distinguish one entity from another entity, without necessarily implying any actual relationship or order between such entities.
[0015]
Embodiments of the present disclosure described herein provide a method and system for segmenting human skin color regions on an image.
[0016]
FIG. 1 is a block diagram of an electronic device 105 for segmenting human skin color regions on an image, in accordance with one embodiment. Examples of the electronic device 105 include, but are not limited to, computer, laptop, mobile device, hand held device, and personal digital assistant (PDA).
[0017]
The electronic device 105 includes a bus 110 for communicating information, and a processor 115 coupled with the bus 110 for processing information. The electronic device 105 also includes a memory 120, for example a random access memory (RAM) coupled to the bus 110 for storing information required by the processor 115. The memory 120 can be used for storing temporary information required by the processor 115. The electronic device 105 further includes a read only memory (ROM) 125 coupled to the bus 110 for storing static information required by the processor 115. A storage unit 130, for
5/20
example a magnetic disk, hard disk or optical disk, can be provided and coupled to bus 110 for storing information.
[0018]
The electronic device 105 can be coupled via the bus 110 to a display 135, for example a cathode ray tube (CRT) or liquid crystal display (LCD), for displaying information. An input device 140, including various keys, is coupled to the bus 110 for communicating information to the processor 115. In some embodiments, cursor control 145, for example a mouse, a trackball, a joystick, or cursor direction keys for communicating information to the processor 115 and for controlling cursor movement on the display 135 can also be present.
[0019]
In some embodiments, the display 135 may perform the functions of the input device 140. For example, consider a touch screen display operable to receive haptic input. The user can then use a stylus to select one or more portions on the visual image displayed on the touch screen device.
[0020]
In some embodiments, the steps of the present disclosure are performed by the electronic device 105 using the processor 115. The information can be read into the memory 120 from a machine-readable medium, for example the storage unit 130. In alternative embodiments, hard-wired circuitry can be used in place of or in combination with software instructions to implement various embodiments.
[0021]
The term machine-readable medium can be defined as a medium providing data to a machine to enable the machine to perform a specific function. The machine-readable medium can be a storage media. Storage media can include non-volatile media and volatile media. The storage unit 130 can be a non-volatile media. The memory 120 can be a
6/20
volatile media. All such media must be tangible to enable the instructions carried by the media to be detected by a physical mechanism that reads the instructions into the machine.
[0022]
Examples of the machine readable medium includes, but are not limited to, a floppy disk, a flexible disk, hard disk, magnetic tape, a CD-ROM, optical disk, punchcards, papertape, a RAM, a PROM, EPROM, and a FLASH-EPROM.
[0023]
The machine readable medium can also include online links, download links, and installation links providing the information to the processor 115.
[0024]
The electronic device 105 also includes a communication interface 150 coupled to the bus 110 for enabling data communication. Examples of the communication interface 150 include, but are not limited to, an integrated services digital network (ISDN) card, a modem, a local area network (LAN) card, an infrared port, a Bluetooth port, a zigbee port, and a wireless port.
[0025]
In some embodiments, the processor 115 can include one or more processing units for performing one or more functions of the processor 115. The processing units are hardware circuitry performing specified functions.
[0026]
The one or more functions include extracting fuzzy rules corresponding to an image data set based on processing parameters. The processing parameters indicate human skin regions and non- human skin regions associated with the image data set. The one or more functions also include converting the fuzzy rules to radial basis function rules. Further, the one or more functions include applying the radial basis function rules on a visual image and rendering human skin color regions on the visual image in a first
7/20
distinctive color and the non- human skin color regions in a second distinctive color. The one or more functions also include receiving an input associated with at least one of the human skin color regions and the non- human skin color regions from one or more portions on the visual image rendered, altering the processing parameters associated with the one or more portions of the visual image to reassign human skin color regions and non- human skin color regions corresponding to the one or more portions based on the input, and generating additional radial basis function rules based on the altered parameters. Further, the one or more functions include rendering the one or more portions on the visual image in one of the first distinctive color and the second distinctive color based on the additional radial basis function rules.
[0027]
Further, the electronic device 105 includes an image sampler 155 for comparing color values of one or more portions of the visual image with color values in the image data set and assigning the human skin color regions and the non-human skin color regions to the one or more portions based on the processing parameters.
[0028]
The storage unit 130 stores the image data set corresponding to multiple sample images. The image data set includes image data associated with each image sampled.
[0029]
In an embodiment, the processor 115 can include an encoding unit 160 for converting the fuzzy rules to radial basis function rules.
[0030]
FIG. 2 (Prior Art) is an exemplary illustration of a fuzzy decision tree.
[0031]
Fuzzy decision trees are composed of a set of internal nodes representing variables used in the solution of a classification problem, a set of branches representing
8/20
fuzzy sets of corresponding node variables, and a set of leaf nodes representing the degree of certainty with which each class has been approximated. The first internal node of the tree is the root node represented here by Vroot. Each new training pattern is classified by starting from the root node and then reaching to one or more than one leaf nodes by following the path of degree of membership greater than zero.
[0032]
A traversing path from to each leaf node will be represented by:
[0033]
pathk : xk1 is Fk1 ^ xk2 is Fk2 ^ …. ^ xkpk is Fkpk;
[0034]
leafk : y1 = 1(βk1)……; yq = q(βkq ); k = 1…….Mβ
[0035]
Here M is the number of paths. Each pathk is defined on premise space composed of the input vector xkwhere xk = [xk1, xk2,… xpk], Fkj are linguistic labels of membership functions defined on the corresponding input universe of discourse, ‘^’ is a fuzzy T-norm operator, leafk is the degree of certainty (βkl; l = 1,…., q), with which pathk assigns a given pattern to class l.
[0036]
FIG. 3 (Prior Art) is an exemplary illustration of generalized Gaussian radial basis function network.
[0037]
Generalized Gaussian RBF networks Generalized Gaussian RBF network classifier under consideration is defined YL = f(x) =k=1ΣM WkL * Øk(x k) ; L=1 to q, where x∈Rp; p is the number of input variables, xk ∈RPk ; pk < p. y = [y1,..,yq] ∈Rq, can be interpreted as the degree of membership of a training pattern to q classes of concern. The network has M nonlinear processing units and the nonlinearity of the kth unit is represented
9/20
by the function Øk(x k) = exp[-(x k - ck)’ Δ k (x k - ck) where ck belongs to RPk is the center of kth radial basis function unit with a width matrix Δ k belonging to RPk * Pk. The width parameter Δ k is a diagonal matrix, wherein Δkj = 1/(σkj)2 if k=j, otherwise Δkj=0. This way, each non-linear processing unit is a Gaussian type ellipsoidal basis function WkL is the weight from kth RBF unit to Lth classification node. When classification to only one class is required, the class with the highest membership degree has been selected, i.e., classify given pattern to class L0, where L0 = argL=1 max YL. Thick lines in the illustrated figure represent vectors.
[0038]
FIG. 4a – FIG. 4b is a flow chart illustrating a method for segmenting human skin color regions on an image, in accordance with one embodiment.
[0039]
Multiple images are electronically sampled to classify color values corresponding to the images as human skin color regions and non-human skin color regions. An image data set is generated based on the classification and stored. For example, consider ‘N’ images are electronically sampled. Each pixel on an image sampled is defined as a human skin color region or a non-human skin color region by determining red(R), green (G) and blue (B) color values from an RGB color model corresponding to the pixel. Further, each pixel defined as human skin color is assigned a skin color classification ‘class-1’ and each pixel defined as non-human skin color is assigned a skin color classification ‘class-2’.
[0040]
Each pixel of an image is characterized by the R, G, and B color values of the pixel and the skin color classification. Exemplarily, for “X” number of pixels in the image sampled, an image data matrix corresponding to the “X” pixels can be represented
10/20
exemplarily as “X rows* 4 columns”. First three columns correspond to the R, G and B color values of each pixel on the image and last column corresponds to the skin color classification assigned.
[0041]
An image data set includes image data associated with the each image sampled among the “N” images. A Fuzzy decision tree (FDT) corresponding to the image data set is generated. The fuzzy decision tree can be generated on the image data set based on, but not limited to, an induction algorithm. Examples of the induction algorithm include, but are not limited to, fuzzy Iterative Dichotomiser 3 (ID3) algorithm. The fuzzy decision tree includes, but is not limited to, fuzzy rules associated with the image data set.
[0042]
The method starts at step 405.
[0043]
At step 410, fuzzy rules corresponding to an image data set are extracted. The fuzzy rules can be extracted from the fuzzy decision tree. The extraction is based on processing parameters. The processing parameters indicate human skin color regions and non-human skin color regions associated with the image data set. The processing parameters are based on color models and include color values of R, G, and B corresponding to the human skin color and non-human skin colors. The processing parameters, include, but are not limited to, skin color classification associated with each pixel of each image in the image data set,
[0044]
The fuzzy rules include rules for comparison of the RGB color values of one or more portions on a visual image with the RGB color values of the pixels corresponding to each image sampled in the image data set. The one or more portions on the visual image with RGB color values matching with the RGB color values in the image data set can be
11/20
classified as human skin color regions or non-human skin color regions based on the skin color classification assigned corresponding to the RGB color values in the image data set.
[0045]
At step 415, the fuzzy rules are converted to radial basis function (RBF) rules. The conversion involves mapping the fuzzy rules to a Gaussian RBF network (GRBFN). The conversion may be performed by a GRBFN encoder. The GRBFN encoder receives the FDT structure as input and maps it to a generalized GRBFN structure.
[0046]
At step 420, the radial basis function rules are applied on the visual image. The RGB color values corresponding to each pixel on the visual image are compared with the RGB values in the image data set. The radial basis function rules determine if the RGB color value corresponding to a pixel of the visual image matches with the RGB color values in the image data set. The radial basis function rules assign each pixel of the visual image with the skin color classification assigned to the matching RGB color values in the image data set. Each pixel is determined to be a human skin color region or a non-human skin color region based on skin color classification.
[0047]
At step 425, the human skin color regions and non-human skin color regions on the visual image are rendered in a first distinctive color and a second distinctive color respectively. The visual image rendered is herein referred to as a processed image. For example, the human skin color regions may be rendered in white color corresponding to color value (255, 255, 255), and non-human skin color region may be rendered in black color corresponding to the color value (0, 0, 0). The rendering is based on the determination of the skin color classification of the pixels of the visual image as described in step 420. .
12/20
[0048]
At step 430, an input associated with at least one of the human skin color regions and the non- human skin color regions on the processed image is received from the user. The input is indicative of selection of the one or more portions on the processed image as one of human skin color region and non-human skin color region. The selection can be performed by the user by electronically identifying the one or more portions on the processed image.
[0049]
In an embodiment, the selection of the one or more portions on the input visual image may be performed prior to the rendering in step 425.
[0050]
At step 435, the processing parameters associated with the one or more portions of the processed image are altered to reassign human skin color regions and non- human skin color regions from the one or more portions on the visual image. The altering can correspond to intelligent learning by the device. The altering is based on the user input of the selection of the one or more portions corresponding to the one or more portions of the processed image.
[0051]
The RGB color values corresponding to the selected portions on the visual image are identified in response to the input. Subsequently, matching color values of the selected portions with the color values in the image data set, the skin color classification is assigned.
[0052]
In some embodiments, the processing parameters can be altered using a gradient descent algorithm. The altering may include setting a fixed learning rate, and error goal for improving the accuracy of classifying human skin color regions and non-human skin color regions. 13/20
[0053]
At step 440, additional RBF rules are generated based on the altered parameters. The additional RBF rules take into account the altered parameters and skin color classification associated with the one or more portions. The additional rules improve accuracy of displaying the one or more portions on the processed image in skin color classification (human or non human) newly determined for consequent application of the RBF rules on the processed image again. Upon generating the additional RBF rules, the additional RBF rules are applied on the processed image as described in step 420.
[0054]
The one or more portions on the processed image are rendered in one of the first distinctive color and the second distinctive color based on additional radial basis function rules herein known as the rendered image. The first distinctive color and the second distinctive color correspond to human skin color regions and non-human skin color regions respectively.
[0055]
For example, consider a portion of the image identified as non-human skin color region on the processed image. If the user selects the portion as human skin-color region, the RBF color value corresponding to the portion is identified. The category of the RBF color value within the GRBFN structure, corresponding to the portion is changed to match the user selection. During the rendering, the visual image renders the selected portion in the first distinctive color.
[0056]
At step 445, the step ends.
[0057]
FIG. 5 is an exemplary illustration of segmenting human skin color regions on an image, in accordance with one embodiment.
14/20
[0058]
Consider a visual image 505 for segmenting human skin color regions and non-human skin color regions. The visual image 505 is processed for highlighting the human skin color regions and non-human skin color regions on the visual image in white color and black color respectively. In the illustration, the colors of the hockey stick in the visual image 505, is similar to the human skin-color region. As a result, at the end of the image processing the hockey stick is represented as white colored hockey stick in the processed image 510. The user may correct the error by selecting a portion corresponding to the hockey stick and indicate that the portion corresponding to the hockey stick is a non-human skin color region. The visual image 505 is processed and rendered again with the selected portion highlighted in colors corresponding to the human skin color or non-human skin color based on user input.
[0059]
In the preceding specification, the present disclosure and its advantages have been described with reference to specific embodiments. However, it will be apparent to a person of ordinary skill in the art that various modifications and changes can be made, without departing from the scope of the present disclosure, as set forth in the claims below. Accordingly, the specification and figures are to be regarded as illustrative examples of the present disclosure, rather than in restrictive sense. All such possible modifications are intended to be included within the scope of the present disclosure.
15/20
I/We claim:
1.
A method for segmenting human skin color regions on an image, the method comprising:
extracting fuzzy rules corresponding to an image data set based on processing parameters, wherein the processing parameters indicate the human skin color regions and non-human skin color regions associated with the image data set; and
converting the fuzzy rules to radial basis function rules.
2.
The method of claim 1, wherein the processing parameters are based on color models.
3.
The method of claim 1, wherein the image data set comprises color values associated with multiple image samples.
4.
The method of claim 1 further comprising:
applying the radial basis function rules on a visual image; and
rendering the human skin color regions on the visual image in a first distinctive color and the non- human skin color regions on the visual image in a second distinctive color based on the applying.
5.
The method of claim 4, wherein the applying comprises:
16/20
comparing color values of one or more portions of the visual image with color values in the image data set; and
assigning human skin color regions and non-human skin color regions to the one or more portions based on the processing parameters..
6.
The method of claim 4 further comprising:
receiving an input associated with at least one of the human skin color regions and the non- human skin color regions from one or more portions on the visual image rendered;
altering the processing parameters associated with the one or more portions of the visual image to reassign human skin color regions and non- human skin color regions corresponding to the one or more portions based on the input; and
generating additional radial basis function rules based on the altered parameters.
7.
The method of claim 6 further comprising:
rendering the one or more portions on the visual image in one of the first distinctive color and the second distinctive color based on the additional radial basis function rules.
8.
An electronic device for segmenting human skin color regions on an image comprising: 17/20
a communication interface that receives an input associated with at least one of the human skin color regions and non- human skin color regions from one or more portions on a visual image;
a memory for storing information; and
a processor responsive to the information to
extract fuzzy rules corresponding to an image data set based on processing parameters, wherein the processing parameters indicate the human skin color regions and the non-human skin color regions associated with the image data set; and
convert the fuzzy rules to radial basis function rules.
9.
The electronic device of claim 8, wherein the processor is further operable to perform:
apply the radial basis function rules on a visual image; and
render the human skin color regions on the visual image in a first distinctive color and the non- human skin regions in a second distinctive color.
10.
The electronic device of claim 9 further comprising:
an image sampler to:
compare color values of one or more portions of the visual image with color values in the image data set; and
assign the human skin color regions and the non-human skin color regions to the one or more portions based on the processing parameters.
18/20
11.
The electronic device of claim 9, wherein the processor is further operable to perform:
receive an input associated with at least one of the human skin color regions and the non-human skin color regions from the one or more portions on the visual image rendered;
alter the processing parameters associated with the one or more portions of the visual image to reassign the human skin color regions and the non- human skin color regions corresponding to the one or more portions based on the input; and
generate additional radial basis function rules based on the altered parameters.