CN102521867B - Mobile phone anime character and background creation method - Google Patents

Abstract

The invention discloses a mobile phone animation character and background creation method, which uses any one of the following five methods to process the original image or uses any one of the first 4 methods and method 5 to mix and process the source image : Method 1: Off-diagonal element blanking method; Method 2: Random blurring method; Method 3: Bead curtain effect simulation method; Method 4: Image color shifting method; Method 5: Monochrome and passivation method. The method is novel in conception, not only interesting, but also simple in implementation, small in data volume, easy to implement, and suitable for running on mobile phones.

Description

A mobile phone animation character and background creation method

technical field

The invention relates to a mobile phone animation character and background creation method.

Background technique

As a branch of the animation industry, my country's mobile animation products have also achieved considerable development. Currently, they mainly include animation MMS, four-frame mobile comics, animation short films and turn-based mobile games. They have the dual characteristics of both cultural products and information products: as cultural products, it is undoubtedly necessary to follow the general principles of socialist cultural products, but it is not suitable for blunt preaching, but to entertain and teach; as information products, we hope to introduce as much as possible Advanced computer technology, in order to improve product quality and improve its production efficiency.

Obscurity in art is an aesthetic realm. Compared with clear and distinct true-color digital works, it seems to be an art form with special aesthetic feeling that can only be understood but hard to describe. If the true-color digital photos are compared to the fine brushwork flowers and birds in traditional Chinese paintings, and the ladies with heavy makeup, then all kinds of hazy works are like the splashed ink landscapes in Chinese paintings, and they are village girls who do not wear makeup. Shino: "The water is shining It’s good to have a sunny day, but it’s strange to see the mountains empty and rainy.” Compared with colorful and vivid color digital photos, the faintly visible and empty mountain scenery pursued by the present invention will leave infinite imagination space for the audience.

Have not yet found the report of this kind of animation works similar to the present invention with the effects of color-gathering or black-and-white hazy paintings.

Contents of the invention

The technical problem to be solved by the present invention is to provide a mobile phone cartoon character and background creation method. The mobile phone cartoon character and background creation method of the present invention has a small amount of data processing, is easy to implement, and is suitable for running on a mobile phone.

The technical solution of the invention is as follows:

A mobile phone animation character and background creation method, using any one of the following five methods to process the original image or using any one of the first 4 methods and method 5 to mix and process the source image:

Method 1: Off-diagonal element blanking method:

Method 2: random fuzzy method;

Method 3: Simulation method of bead curtain effect;

Method 4: image color shifting method;

Method 5: monochromatization and passivation method.

The steps of the described non-diagonal element blanking method are:

The original image is gridded, and the pixels at the diagonal of each grid retain the color, and the pixels at other positions are cleared to white, that is, replaced with white pixels;

The grid density corresponding to gridSize is less than 5 pixels; the grid density is the side length of the square grid.

The steps of the random fuzzing method in Method 2 are:

Step (1). The data space reserved for storing the output image, that is, the target image, is cleared to white;

Step (2). Starting from the upper left corner of the original image, parameter i is a row counter, parameter j is a column counter, both parameter i and parameter j use gridSize as the incremental step, parameter k is an auxiliary counter, and k is incremented from 0 to gridSize , and then reset to zero after reaching the gridSize; traverse the whole graph according to steps (3)~(4), that is, complete the random fuzzing process;

Step (3). In each grid with gridSize as the side length, determine the source and final two types of pixel points, namely pixelSou and pixelDes positions, as follows:

pixelSou=souBase+(i+k+yrand)*lineByte+(j+k+xrand)*pixelByte;

pixelDes=desBase+(i+k)*lineByte+(j+k)*pixelByte;

Among them, souBase and desBasewe are the base addresses of the source and final pixel points respectively, lineByte is the number of bytes per line of the image, pixelByte is the number of bytes per pixel of the image, i and j are the serial numbers of the corresponding rows and columns, yrand and xrand are both It is a random number, and the random number takes a random value between 0-gridsize;

Step (4). Take out the blue, green, and red data of the source pixel and assign it to the corresponding output pixel

*(pixelDes) = *(pixelSou);

*(pixelDes+1)=*(pixelSou+1);

*(pixelDes+2)=*(pixelSou+2).

The bead curtain effect simulation method is:

Step 1: Make the bead curtain template: set the bead curtain parameters to make the bead chain template. The bead curtain parameters include the row spacing of the bead chain, the row spacing of the bead body, the diameter of the bead body, and the diameter of the connecting line of the bead body [In the example, the connection line Diameter is 2]

Step 2: Superimpose the bead curtain template as a mask on the image formed by method 1 or method 2 to obtain a final new image with bead curtain effect.

The steps of the image color shifting method are:

Step (1). Select 2 color sample points, and according to the positions of the 2 color sample points, find out the blue, green, and red values corresponding to their pixels from the image, respectively denoted as b1, g1, r1, and b2 , g2, r2;

Step (2). The initial coordinates of the processing point X and Y are set to the upper left corner of the image, that is, both X and Y are set to 0;

Step (3). Through the following steps (4) to (9), proceed column by column and row by pixel, that is, process points:

Step (4). Obtain the blue, green and red values of the current processing point, denoted as sb, sg, sr respectively;

Step (5). If abs(sb-b1)<areaExtanDegree and abs(sg-g1)<areaExtanDegree and abs(sr-r1)<areaExtanDegree, record b1 as b, record g1 as g, record r1 as r, and go to step (7) Carry out image color shifting operation; among them, areaExtanDegree is the degree of area expansion; otherwise, if the judgment condition is not satisfied, go to step (8) to retain the original color; [color shifting operation will bring the effect of mountain rain: because of mountain rain In the future, the sample point is taken in the sky (or mountain), and this judgment is to measure the range of the sky (or mountain). Because in the RGB color standard, the smaller the RGB value, the darker the color. For example, when the RGB three colors are all 0, it is black. The essence of the color shifting operation in step 7 is to reduce the value of RGB, so as to darken the sky (or the color of the mountain) and achieve the effect of rain coming]

Step (6). If abs(sb-b2)<areaExtanDegree and abs(sg-g2)<areaExtanDegree and abs(sr-r2)<areaExtanDegree, record b2 as b, record g2 as g, record r2 as r, and go to step (7) Carry out image color shifting operation; On the contrary, if discriminating condition does not satisfy then go to step (8) retain primary color;

Step (7). Transform the RGB system of the color into the YCrCb system according to the following formula

Y＝(257*r+504*g+98*b)/1000+16;

Cr＝(439*r-368*g-71*b)/1000+128;

Cb=(-148*r-291*g+439*b)/1000+128;

Correct the Y value according to the fog concentration parameter:

Y=Y*fog;

Convert the YCrCb system to the RGB system inversely according to the following formula:

fogB=(1.164*(Y-16)+2.017*(Cb-128));

fogG=(1.164*(Y-16)-0.813*(Cr-128)-0.392*(Cb-128));

fogR=(1.164*(Y-16)+1.596*(Cr-128));

Dim and yellow the current pixel value and output

*(pixelDes)=sb-(fogB*fog/10);

*(pixelDes+1)=sg-(fogG*fog/10);

*(pixelDes+2)=sr-(fogR*fog/10);

Go to step (9);

Step (8). Keep the image data of the processing point as the original data output:

*(pixelDes) = sb;

*(pixelDes+1)=sg;

*(pixelDes+2)=sr;

Step (9). The X coordinate of the processing point is +1, if the right boundary of the image is not reached, then go to step (4); otherwise go to step (10);

Step (10). The X coordinate of the processing point is set to 0, and the Y coordinate of the processing point is +1. If the processing point does not reach the lower right boundary of the image, go to step (4); otherwise, the program ends.

Described monochromaticization and passivation method are:

The monochrome operation is first performed on the original image, that is, the following grayscale operation is performed on each pixel of the image:

Gray value I=0.3B+0.59G+0.11R of each pixel after processing, wherein B, G, R are the original blue, green and red component values of the pixel; obtain the monochromatic image;

Then perform a passivation operation on the monochromatic image; the passivation operation is as follows, for an N*N filter window, calculate the maximum value, median value and minimum value for each column element of the filter window; then find all The minimum value MaxMin among the maximum values, the median value MedMed among all intermediate values, and the maximum value MinMax among all minimum values, the finally obtained median value is MED=med[MaxMin, MedMed, MinMax].

Because no one has thought of using photos of oneself (or friends) as materials on mobile phones to carry out DIY operations to create mobile animation characters or animation backgrounds; especially considering that mobile phones are far weaker than desktop computers (only equivalent to this century) The computing power of early desktop computers) is also often prohibitive.

In order to realize DIY image operation for the special platform of mobile phone, the following series of measures are adopted in the design, especially to reduce the memory requirement of the application program:

A. The simpler the application, the better. We try our best to make each component into a Midlet, and encapsulate multiple Midlets used in a Midlet package, which enables the program manager of the mobile phone to manage Midlets and the resources used by Midlets more economically.

B. The smaller the application, the better. Delete the temporarily unused components in the application and minimize unnecessary information to reduce the size of the entire program. When downloading apps over a wireless network, a smaller app will greatly reduce download time and will run in compatibility (rather than exclusive) with other apps on the device.

C. Minimize the total memory requirements of the application. The main measures are: ①Use less object types and use scalar types instead. Because the scalar type occupies less memory than the object type; ② declare as few objects as possible. Because the system allocates space on the running heap when declaring an object, it should be allocated when the application is about to use the object, rather than all allocated when the program starts. Moreover, once the program no longer needs the object, all references to the object are assigned the value null. ③According to the accuracy needs to use the data type. Whenever possible, data types such as boolean, byte, and short should be used instead of int. This kind of detail has little impact on desktop programs, but it will have a cumulative impact on mobile phones. ④Reuse as much as possible. Let multiple references use the same object at different times in the program's lifetime. For example reusing some large arrays, reusing can take advantage of allocated runtime memory, using "lazy" instantiation. While this doesn't conform to software engineering principles, it fits the realities of a computing device as thin as a cell phone. ⑤ Avoid creating objects within the loop. ⑥ Always check the memory usage. Related methods are: freeMemory and totalMemory. Handle OutMemoryError errors yourself. It should be ensured that when the application runs out of memory, there is a predetermined exit routine to manage this, rather than leaving it to the operating system. ⑦Release resources in time. For resources such as files, network connections, etc., when you no longer need to use them, don't hold them. You should perform the necessary cleanup yourself, rather than relying on the garbage collector or the host environment. ⑧ Use local variables more. In desktop applications, developers are accustomed to setting more class data members and using less local variables. However, class data members are actually "global variables" in the class, which require frequent data scheduling and stack operation support, and actually consume CPU calculations to support them. You can reduce your application's CPU processing by eliminating the extra step of accessing data members of a class through local variable assignment. In this way, although the benefits of encapsulating data in classes are lost, the processing speed needs to be considered first for applications that run on tiny computing devices such as mobile phones and require a large amount of data.

Beneficial effect:

The method for creating mobile cartoon characters and backgrounds of the present invention adopts image processing algorithms, uses digital photos as materials, transforms them into different color registers or black and white hazy paintings, and performs DIY operations on animation servers or personal mobile phones of mobile phone users to create effects Unique anime characters or anime backgrounds.

The present invention uses a simple and practical digital image processing algorithm, pays attention to reducing the amount of data and improving the operating efficiency of the CPU, so that mobile animation works that are restricted by the size of the mobile phone screen and the computing power of the mobile phone can be produced for different users with low creation costs and fast running effects. Mobile phone users of different ages, different levels, and in different time periods can provide electronic information in different styles that are both solemn and harmonious, and bring more beauty and joy to the lives of mobile phone users. In particular, it should be pointed out that one of the purposes of the present invention is to realize DIY image operation on a platform such as a mobile phone whose hardware and software resources are far weaker than that of a computer, so it is absolutely necessary to take the above-mentioned series of targeted measures in the design.

Description of drawings

Fig. 1 is a schematic diagram of the off-diagonal element blanking method; (the step size step in Fig. 1 is gridSize),)

Fig. 2 The original image of the digital photo and the effect images obtained by using five methods (where (a) to (d) are the original images, (e) and (f) are the effect images using the off-diagonal element blanking method ( It can also be vividly called "Grassy Manman" rendering), (g), (h) are renderings using random fuzzing method (also can be vividly called random hazy rendering), (i), ( j) is the effect diagram processed by the bead curtain effect simulation method (it can also be referred to as the "bead curtain swaying" effect diagram, (k), (l) is the effect diagram using the image color shifting method, (it can also be visualized It is called the "mountain and rain is coming" rendering, (m), (n) are renderings using monochromatic and passivation methods, (it can also be vividly called "splashing ink freehand brushwork" rendering).

Detailed ways

The present invention will be described in further detail below in conjunction with accompanying drawing and specific embodiment:

The present invention provides five artistic creation methods such as "light gauze", "random haze", "bead curtain swaying", "mountain rain is coming" and "splashing ink freehand". Among them, the creative method of "light gauze" is realized by the original "non-diagonal element blanking" algorithm; the creative method of "random hazy" superimposes random pixel output on the basis of blanking; the creative method of "bead curtain swaying" The bead curtain effect simulation is superimposed on the basis of blanking; while the basis of "splash ink freehand brushwork" is monochromatic and passivation. However, the creative method of "Mountains and Rain is Coming" has not been eliminated, but an original image color shifting algorithm has been applied, and it is expressed by the following definition 2.

Definition 1. If an image is divided into several grids, and some pixels in each grid are hidden according to a certain rule, it is called image blanking. Image blanking can achieve the effect of gauze mist and looming. In particular, when only the image elements on the diagonal are kept, it is called depth blanking or off-diagonal element blanking

(See Figure 1); On the contrary, if the image elements on off-diagonal lines are kept, it is called shallow blanking or diagonal element blanking.

The looming visual effect can be realized by image blanking technology (especially depth blanking). For example, when the length and width of the grid are both gridSize, if deep blanking is used, only gridSize pixels on the diagonal line will be reserved; if shallow blanking is used, gridSize*(gridSize-1 pixels on the off-diagonal line will be kept ) pixels;

Definition 2. If the color of a pixel is corrected to its original value minus the product of it and a non-negative (or non-positive) coefficient, it is called image color shifting for the pixel. In particular, when a non-negative coefficient is selected, it is called a yellowish color shift; when a non-positive coefficient is selected, it is called a whitening color shift. specifically:

A. The original brightness Y of the pixel is given by the following formula:

Y＝0.257*r _in +0.504*g _in +0.98*b _in (1)

Among them _, r _in , g _in , and bin are the original color values of the pixel.

B. The color shifting coefficient M is an effective step-type non-negative (or non-positive) coefficient that is negatively correlated with the brightness Y.

C. color shifting output r _out , g _out , b _out are given by the following formula;

r _out ＝abs(r _in -(M ₁ *r _in )) (2)

g _out ＝abs(g _in -(M ₂ *g _in )) (3)

b _out ＝abs(b _in -(M ₃ *b _in )) (4)

In the formula, the abs function is an absolute value operation.

Example 1:

To implement the mobile cartoon character and background creation method of the present invention, at first the item parameters will be set as follows:

1. The density of the gauze curtain is the grid density (gridSize): the density of the gauze curtain should be <5.

2. Fog concentration (fog): The fog concentration should be between 1.0 and 2.0.

3 Color Samples (colorSamples): The user selects 2 color sample points through WYSIWYG on-site, and the program atomizes the atomized area determined by the area expansion degree, resulting in a hazy visual effect. Moreover, the color of the bead curtain is made according to the colors of the two color sample points, so the color of the bead curtain is flexible and changeable.

4. AreaExtanDegree (areaExtanDegree): The larger the selected value, the larger the color shifting area extended through the color hazy template point. The optional range is 1 to 100.

For hazy paintings with different backgrounds and characters, see Figure 2.

Method 1: "Blank off-diagonal elements" method, that is, first clear the data space reserved for storing the output image to white, and then hide the image elements on the off-diagonal line, that is, only the Image elements on the corners are left. Specifically, order the row counter and the column counter to use gridSize [it has been set before when setting parameters: the screen density (gridSize): the screen density should be <5. 】Traversing the whole image by "jumping" for the step length, and when the row counter or column counter jumps one step, because another grid is started, the auxiliary counter changes from 0 to gridSize, so that only the image on the diagonal of the source image Copy it to the target image. Through this algorithm of leaving a few and hiding the majority, the effect of light gauze and mist is achieved. Figure 1 is a diagram of the "off-diagonal element blanking" algorithm.

Step (1). The data space reserved for storing the output image is cleared to white;

Step (2). Starting from the upper left corner of the image, parameter i is a row counter, parameter j is a column counter, gridSize is a step size, and parameter k is an auxiliary counter, and traverse the entire image according to steps (3) to (4);

Step (3). Determine the position of the source (source), target (destination) two types of pixel points (pixelSou and pixelDes):

pixelSou=souBase+(i+k)*lineByte+(j+k)*pixelByte;

pixelDes=desBase+(i+k)*lineByte+(j+k)*pixelByte;

Among them, souBase and desBasewe are the base addresses of the source and final pixel points, lineByte is the number of bytes per row of the image, pixelByte is the number of bytes per pixel of the image, i and j are the corresponding rows and columns, and k varies from 0 to gridSize ;

The data space reserved for storing the output image is cleared to white; so if no data is output to this pixel, it is white, thus producing a blanking effect.

It can be understood that the entire image is gridded, the pixels at the diagonal of the grid are reserved, and the pixels at other points are set to white.

Method 2: Implement random blurring. The basis of the stochastic fuzzing method is also the "off-diagonal element blanking" algorithm. Its characteristic is that the row counter and column counter still "jump" through the whole graph with the gridSize as the step size, but add random numbers to the auxiliary counters that affect the source address. in,

(1) The row counter and column counter that constitute the source pointer step by step, the auxiliary counter has 2 variables, one is k (change from 0 to gridSize), and the other is random number yrand or xrand [the range of random number is 0-gridsize Random between] to copy the source image at the specified location to the target image;

(2) The row counter and column counter that constitute the target pointer leap forward step by step, and the auxiliary counter has only one variable k (changed from 0 to gridSize).

Step (1). The data space reserved for storing the output image is cleared to white;

Step (2). Starting from the upper left corner of the image, parameter i is a row counter, parameter j is a column counter, parameter k is an auxiliary counter, gridSize is used as a step size, and traverse the entire image according to steps (3) to (4);

Step (3). Determine the source (source) and final (destination) two types of pixel points (pixelSou and pixelDes) positions:

pixelSou=souBase+(i+k+yrand)*lineByte+(j+k+xrand)*pixelByte;

pixelDes=desBase+(i+k)*lineByte+(j+k)*pixelByte;

Among them, souBase and desBasewe are the base addresses of the source and final pixel points, lineByte is the number of bytes per row of the image, pixelByte is the number of bytes per pixel of the image, i and j are the corresponding rows and columns, and k varies from 0 to gridSize , both yrand and xrand are random numbers;

Step (4). Take out the blue, green, and red data of the source pixel and assign it to the corresponding output pixel

*(pixelDes) = *(pixelSou);

*(pixelDes+1)=*(pixelSou+1);

*(pixelDes+2)=*(pixelSou+2).

Method 3: Realize the bead curtain effect. The basis of the bead curtain effect simulation method is also the "off-diagonal element blanking" algorithm. The row counter and column counter "jump" through the whole graph with gridSize as the step size. Wherein, when the row counter and the column counter take a step, the auxiliary counter changes from 0 to gridSize, thereby blanking off-diagonal elements in the source image. (see Fig. 1) and, because be to make the color of bead curtain according to the color of 2 color sample points, so have very strong flexibility.

Step (1). The data space reserved for storing the output image is cleared to white;

Step (2). Starting from the upper left corner of the image, the parameter i is a row counter, the parameter j is a column counter, the parameter k is an auxiliary counter, gridSize is used as a step size, and traverse the whole image according to steps (2) to (4);

Step (3). Determine the source (source), target (destination) two types of pixel points (pixelSou and pixelDes) position:

pixelSou=souBase+(i+k)*lineByte+(j+k)*pixelByte;

pixelDes=desBase+(i+k)*lineByte+(j+k)*pixelByte;

Among them, souBase and desBasewe are the base addresses of the source and final pixel points, lineByte is the number of bytes per row of the image, pixelByte is the number of bytes per pixel of the image, i and j are the corresponding rows and columns, and k varies from 0 to gridSize ;

Step (4). Take out the blue, green, and red data of the source pixel and assign it to the corresponding output pixel

*(pixelDes)=abs(*(pixelSou));

*(pixelDes+1)=abs(*(pixelSou+1));

*(pixelDes+2)=abs(*(pixelSou+2));

Step (5). According to the positions of the two color template points, their pixel data color1 and color2 are respectively found from the image;

Step (6). Use two colors of color1 or color2 with appropriate row spacing and column spacing [for example, the row spacing in Figures 2i and 2j is 20 for each of the two colors alternately, and the column spacing is 40. 】Call the MakeColorPearl template to form colored pearls. The parameters of the template are: column number j, line number i, lineByte of each line of the image, pixelByte of each pixel of the image, pearl diameter r, pearl color (color1 or color2 respectively); [MakeColorPearl template, namely Forming a color bead curtain template is an existing mature technology,]

Method 4: Image color shifting method, comprising the following steps:

Step (1). According to the positions of the two color sample points, find out their pixel data (blue, green, red values) respectively from the image, which are respectively recorded as b1, g1, r1, and b2, g2, r2, and use In step (5) and step (6).

Step (2). The initial coordinates of the processing point X and Y are set to the upper left corner of the image, that is, both X and Y are set to 0;

Step (3). Advance column by column and row by row through the following steps (4) to (9):

Step (4). Obtain the blue, green and red values of a certain processing point, denoted as sb, sg, sr respectively;

Step (5). If abs(sb-b 1)<areaExtanDegree and abs(sg-g1)<areaExtanDegree and abs(sr-r1)<areaExtanDegree, record b1 as b, record g1 as g, record r1 as r, and turn to Step (7) performs image color shifting operation. On the contrary, if the discrimination condition is not satisfied, then go to step (8) to retain the original color;

Step (6). If abs(sb-b2)<areaExtanDegree and abs(sg-g2)<areaExtanDegree and abs(sr-r2)<areaExtanDegree, record b2 as b, record g2 as g, record r2 as r, and go to step (7) Carry out image color shifting operation. On the contrary, if the discrimination condition is not satisfied, then go to step (8) to retain the original color;

Step (7). According to formula (1), (2), (3) the RGB system of color is transformed into YCrCb system

Y＝(257*r+504*g+98*b)/1000+16 (5)

Cr＝(439*r-368*g-71*b)/1000+128 (6)

Cb＝(-148*r-291*g+439*b)/1000+128(7)

Correct the Y value according to the fog concentration parameter:

Y＝Y*fog (8)

According to the formula (5), (6), (7) inversely convert the YCrCb system to the RGB system

fogB＝(1.164*(Y-16)+2.017*(Cb-128)) (9)

fogG＝(1.164*(Y-16)-0.813*(Cr-128)-0.392*(Cb-128)) (10)

fogR＝(1.164*(Y-16)+1.596*(Cr-128)) (11)

Dim and yellow the current pixel value and output

*(pixelDes)=sb-(fogB*fog/10);

*(pixelDes+1)=sg-(fogG*fog/10);

*(pixelDes+2)=sr-(fogR*fog/10);

Go to step (9);

Step (8). Output the image data of the processing point "as it is":

*(pixelDes) = sb;

*(pixelDes+1)=sg;

*(pixelDes+2)=s r;

Step (9). The X coordinate of the processing point is +1, if the right boundary of the image is not reached, then go to step (4); otherwise go to step (10);

Step (10). The X coordinate of the processing point is set to 0, and the Y coordinate of the processing point is +1. If the processing point does not reach the lower right boundary of the image, then go to step (4); otherwise, the program ends;

Algorithm 5. Monochrome and passivation method: call grayscale operation and image passivation operation successively. Grayscale operation, the grayscale value of each pixel after processing is I=0.3B+0.59G+0.11R, where B, G, and R are the original blue, green, and red component values of the pixel.

Passivation methods are described below:

The standard one-dimensional median filter is defined as:

y _k ＝med{x _kn ，x _k-n+1 ，...，x _k ，...，x _k+n-1 ，x _k+n }

In the formula, med means to take the median value operation. It can be seen that the median filter is to convert the grayscale of each pixel of the sliding window (that is, x _kn , x _k-n+1 , ..., x _k , ..., x _k+n-1 in the above formula , x _k+n ) are sorted from top to bottom, and the middle value is selected as the representative of the pixel. When the sliding window is large and there are many pixels in the sliding window (set to m), it is very computationally expensive to perform a sorting operation (need to do m(m-2)/2 comparison operations. A quick sorting algorithm is used for this purpose. The passivation operation reduces the calculation amount by about 0.47 times. The idea is briefly described as follows: for the convenience of description, the pixels in each row in the 3×3 window are called: row 0 p _.0 , p ₁ , p ₂ , row 1 p _{. 3} , p _.4 , p ₅ , the second line p ₆ , p _.7 , p _.8 . First, calculate the maximum value group, median value group, and minimum value group for each column of pixels in the window:

Maximum group: Max ₀ = max[p ₀ , p _.3 , p ₆ ], Max ₁ = max[p ₁ , p ₄ , p ₇ ], Max ₂ = max[p ₂ , p ₅ , p ₈ ]; Similarly there are median groups Med ₀ , Med ₁ , Med ₂ , and minimum groups Min ₀ , Min ₁ , Min ₂ . Further obtain: MaxMin=min[Max ₀ , Max ₁ , Max ₂ ], similarly obtain MedMed, MinMax; further obtain: MED=med[MaxMin, MedMed, MinMax] is the real median value.

Claims (1)

1. cartoon character on mobile phone and a background creative method, is characterized in that, adopts any method in front 4 kinds of methods and method 5 to source images hybrid processing:

Method 1: off diagonal element blanking method:

Method 2: Random-fuzzy method;

Method 3: pearl-decorated curtain effect simulation method;

Method 4: image moves color method;

Method 5: monochromatization and passivating method;

The step of described off diagonal element blanking method is:

By original image gridding, the pixel at the diagonal line place of each grid retains color, and the pixel zero clearing of other positions is white, replaces by white pixel;

Mesh-density gridSize corresponding to gridding is for being less than 5 pixels; Described mesh-density is the length of side of square net;

Random-fuzzy method step in method 2 is:

Step (1). be that the data space zero clearing of target image is white by the reserved output image of depositing;

Step (2). from the upper left corner of original image, parameter i is linage-counter, and parameter j is column counter, parameter i and parameter j are all using gridSize as incremental steps, parameter k is auxiliary counter, and k is incremented to gridSize from 0, reaches zero clearing again after gridSize; By step (3)～(4) traversal full figure, complete the process of Random-fuzzy;

Step (3). in the grid at each taking gridSize as the length of side, determine that by the following method source, whole two class pixels are pixelSou and pixelDes position:

pixelSou＝souBase+(i+k+yrand)*lineByte+(j+k+xrand)*pixelByte；

pixelDes＝desBase+(i+k)*lineByte+(j+k)*pixelByte；

Wherein souBase and desBase are respectively the base address of source, whole two class pixels, lineByte is the every row byte number of image, pixelByte is the every pixel words joint number of image, i and j are the ordinal number of corresponding line and row, yrand and xrand are random numbers, and random number is random value between 0-gridsize;

Step (4). taking-up source pixel is filled enamel, green, red data are given corresponding output pixel point

*(pixelDes)＝*(pixelSou)；

*(pixelDes+1)＝*(pixelSou+1)；

*(pixelDes+2)＝*(pixelSou+2)；

Pearl-decorated curtain effect simulation method is:

Step 1: make pearl-decorated curtain template: set pearl-decorated curtain parameter and make pelletron template, described pearl-decorated curtain parameter comprises pelletron row distance, pearl body line-spacing, pearl body diameter, pearl body connecting line diameter;

Step 2: described pearl-decorated curtain template is superimposed upon on the image that method 1 or method 2 form and obtains the final new images with pearl-decorated curtain effect as covering plate;

The step that image moves color method is:

Step (1). choose 2 color typical applications, according to the position of 2 color typical applications, from image, find out respectively their corresponding indigo plant, green, the red numerical value of pixel, be denoted as respectively b1, g1, r1, and b2, g2, r2;

Step (2). process points x, y initial coordinate is made as the image upper left corner, i.e. x, y is all made as 0;

Step (3). be according to pixels that process points advances by column, line by line by following steps (4)～(9):

Step (4). obtain the indigo plant of current process points, green, red numerical value, be denoted as respectively sb, sg, sr;

Step (5). as abs (sb-b1) <areaExtanDegree and abs (sg-g1) <areaExtanDegree and abs (sr-r1) <areaExtanDegree remember that b1 is b, note g1 is g, note r1 is r, goes to step (7) and carries out image and move look operation; Wherein, areaExtanDegree is area extension degree; Otherwise, go to step (8) reservation primary colors as criterion does not meet; In formula, abs function is the operation that takes absolute value;

Step (6). as abs (sb-b2) <areaExtanDegree and abs (sg-g2) <areaExtanDegree and abs (sr-r2) <areaExtanDegree remember that b2 is b, note g2 is g, note r2 is r, goes to step (7) and carries out image and move look operation; Otherwise, go to step (8) reservation primary colors as criterion does not meet;

Step (7). as follows the RGB standard transformation of color is become to YCrCb standard

Y＝(257*r+504*g+98*b)/1000+16；

Cr＝(439*r-368*g-71*b)/1000+128；

Cb＝(-148*r-291*g+439*b)/1000+128；

According to fog concentration parameter correction Y value:

Y=Y*fog; Wherein, fog is fog concentration, and value is between 1.0～2.0;

By following formula, the contravariant of YCrCb standard is changed to RGB standard:

fogB＝(1.164*(Y-16)+2.017*(Cb-128))；

fogG＝(1.164*(Y-16)-0.813*(Cr-128)-0.392*(Cb-128))；

fogR＝(1.164*(Y-16)+1.596*(Cr-128))；

Current pixel value is given to dim yellow and move look output

*(pixelDes)＝sb–(fogB*fog/10)；

*(pixelDes+1)＝sg–(fogG*fog/10)；

*(pixelDes+2)＝sr–(fogR*fog/10)；

Go to step (9);

Step (8). the view data of process points is kept to former data output:

*(pixelDes)＝sb；

*(pixelDes+1)＝sg；

*(pixelDes+2)＝sr；

Step (9). process points x coordinate+1, as do not gone to step (4) to image right margin; Otherwise go to step (10);

Step (10). process points x coordinate is made as 0, and process points y coordinate+1, as process points does not go to step (4) to border, image bottom right; Otherwise EOP (end of program);

Described monochromatization and passivating method are:

First original image is carried out to monochromatization operation, each pixel of image is carried out to following gray processing operation:

The gray-scale value I=0.3B+0.59G+0.11R of each picture point after processing, wherein B, G, R are the original indigo plant of this pixel, green, red component value; Obtain the image after monochromatization;

Again the image after monochromatization is carried out to passivation operation; Passivation operation as follows, for the filter window of a N*N, to each column element maximizing, intermediate value and the minimum value of this filter window; Obtain intermediate value MedMed in minimum M axMin, all intermediate values in all maximal values and the maximal value MinMax in all minimum value, the intermediate value finally obtaining is MED=med[MaxMin, MedMed, MinMax again].