CN103200415A - 2D (Two Dimensional) to 3D (Three Dimensional) image processing method - Google Patents

Abstract

The invention discloses a 2D to 3D image processing method. The present invention first obtains two target images, wherein one of the first target images is generated based on the initial 2D image, and the other second target image is generated based on the enlarged 2D image, and the gray scale of the second target image is 0 The second pixel point needs to be repaired by combining the first target image and the second target image to obtain a complete second target image. The second target image is used as the right eye image, and the initial 2D image is used as the left eye image, thus realizing 2D Conversion to 3D.

Description

A 2D to 3D image processing method

technical field

The invention relates to the field of computer video processing, in particular to an image processing method for converting 2D to 3D.

Background technique

Compared with ordinary 2D pictures, 3D is more three-dimensional and realistic, which can make the audience feel immersive. At present, 3D technology can be divided into two types: naked-eye type and glasses type. At present, the naked-eye type 3D technology is mainly used in industrial and commercial display, while the glasses-type 3D technology is mainly used in the consumer market.

The video processing method of 2D to 3D technology usually first obtains the grayscale image of the 2D image, processes the grayscale image to obtain the depth map of the entire 2D image, takes the original image as the left eye image, and then combines the original image and the depth map to generate the right eye image. image. Existing technologies often focus on the generation method of the depth map, and there is less processing of the target image generated based on the original image and the depth map, but in fact, the processing of the target image can make up for the defects of the target image and reduce the generation of the depth map. quality requirements.

Contents of the invention

In view of the above technical problems, the present invention provides a 2D-to-3D image processing method for patching a target image to improve the conversion effect.

The technical scheme provided by the invention is:

An image processing method for converting 2D to 3D, comprising the following steps:

Step 1, generating an initial depth map according to the initial 2D image;

Step 2: Enlarging the initial 2D image by n times to obtain an intermediate image, and enlarging the initial depth image by n times to obtain an intermediate depth image;

Step 3, using the initial 2D image and the initial depth map to obtain the first target image, and using the intermediate image and the intermediate depth map to obtain the second target image;

Step 4: Obtain the coordinates of a second pixel points whose gray scale is 0 in the second target image, and the pixels whose gray scale is not 0 in the second target image are defined as third pixel points;

Step 5. Calculate the coordinates of the first pixel points corresponding to a second pixel points in the first target image. The specific calculation method is: divide the coordinates of a second pixel point by n to obtain a second pixel point The coordinates of a corresponding first pixel point, so as to obtain all the first pixel points corresponding to a second pixel points one-to-one;

Step 6. Obtain the gray scale of all the second pixels. The method for obtaining the gray scale of any second pixel is:

When the grayscale of a first pixel is not 0, assign the grayscale of the first pixel to a corresponding second pixel,

Calculate the first average value of the gray levels of all the third pixels in the circle with the second pixel as the center and rl as the radius in the square area with the largest area, wherein, in the square area, the third pixel The number of points accounts for 90% of the total number of pixels,

Comparing the gray scale of a second pixel with the first average value, when the ratio of the difference between the gray scale and the first average value to the first average value is less than or equal to 0.05, the gray scale is used as the final value of the second pixel point Gray scale, when the ratio of the difference between the gray scale and the first average value to the first average value is greater than or equal to 0.05, calculate all the squares in the circle with the second pixel as the center and r2 as the radius and the largest area The second average value of the gray scale of the third pixel point, wherein the difference between r2 and rl is the width of a third pixel point, and the second average value is used as the final gray scale point of a second pixel point,

When the grayscale of a first pixel is 0, calculate the third average of the grayscales of all the third pixels in the square area with the largest area in a circle with a second pixel as the center and r3 as the radius value, wherein, in the square area, the number of the third pixel points accounts for 95% of the total number of pixels, and the third average value is used as the final gray scale of the second pixel point, wherein r3 is Integer multiple of rl;

Step 7: After reducing all the acquired gray-scale second target images by a factor of n, they are used as right-eye images, and the initial 2D images are used as left-eye images, and the right-eye images and left-eye images are output to the 3D display device.

Preferably, in the image processing method for converting 2D to 3D, n is 4.

Preferably, in the 2D-to-3D image processing method, rl is 2.5 times the width of a third pixel.

Preferably, in the image processing method for converting 2D to 3D, r3 is 2 to 5 times of r1.

The present invention first obtains two target images, wherein one of the first target images is generated based on the initial 2D image, and the other second target image is generated based on the enlarged 2D image, and the gray scale of the second target image is 0 The second pixel point needs to be repaired by combining the first and second purpose images to obtain a complete second purpose image. The second purpose image is used as the right eye image, and the initial 2D image is used as the left eye image, thus realizing 2D Conversion to 3D.

Description of drawings

Fig. 1 is a flowchart of the present invention.

Detailed ways

The present invention will be further described in detail below in conjunction with the accompanying drawings, so that those skilled in the art can implement it with reference to the description.

As shown in Figure 1, the present invention provides a kind of 2D to 3D image processing method, comprises the following steps:

Step 1: Generate an initial depth map according to the initial 2D image.

Step 2: Enlarging the initial 2D image by n times to obtain an intermediate image, and enlarging the initial depth image by n times to obtain an intermediate depth image.

In the present invention, n is preferably 4. The enlargement process is to copy any pixel in the initial 2D image n times and combine it into a pixel in the intermediate image, that is, the area of a pixel in the intermediate image is n times the area of a pixel in the initial 2D image.

Step 3, using the initial 2D image and the initial depth map to obtain the first target image, and using the intermediate image and the intermediate depth map to obtain the second target image.

Step 4: Obtain the coordinates of a second pixel points with a gray scale of 0 in the second target image, and define the pixels with a gray scale other than 0 in the second target image as third pixel points.

When the first target image is generated based on the initial 2D image and the initial depth map, and the second target image is generated based on the intermediate image and the intermediate depth map, pixels with a grayscale of 0 will be formed in the first target image and the second target image, For these pixels with a gray scale of 0, they belong to the defects generated by the image processing process and need to be compensated.

Step 5. Calculate the coordinates of the first pixel points corresponding to a second pixel points in the first target image. The specific calculation method is: divide the coordinates of a second pixel point by n to obtain a second pixel point coordinates of a corresponding first pixel point, so as to obtain all first pixel points corresponding to a second pixel points one-to-one.

The present invention repairs the second target image with the help of the first target image. First, all a first pixel points that have a one-to-one correspondence with a second pixel points are determined. The determination method is through the first pixel point and the second pixel point. As for the coordinates of the pixel point, since the second pixel point is enlarged, the coordinates of the second pixel point should be consistent with the center coordinates of the first pixel point.

Step 6. Obtain the gray scale of all the second pixels. The method for obtaining the gray scale of any second pixel is:

When the gray scale of a first pixel is not 0, assign the gray scale of the first pixel to a second pixel corresponding to it; calculate a circle centered on a second pixel and taking rl as the radius , the first average value of the gray levels of all the third pixels in the square area with the largest area, wherein, in the square area, the number of the third pixels accounts for 90% of the total number of pixels, comparing a The gray scale of the second pixel point and the first average value, when the ratio of the difference between the gray scale and the first average value to the first average value is less than or equal to 0.05, the gray scale is used as the final gray scale of the second pixel point , when the ratio of the difference between the gray level and the first average value to the first average value is greater than or equal to 0.05, calculate all the second pixels in the circle with the second pixel as the center and r2 as the radius in the square area with the largest area. The second average value of the gray scales of three pixels, wherein the difference between r2 and the gate is the width of a third pixel point, and the second average value is used as the final gray scale of a second pixel point.

Based on a certain gray scale acquisition rule, the second purpose image is completely repaired, and a reasonable gray scale acquisition rule can ensure the repair quality of the second purpose image.

Taking a second pixel as an example, when the gray scale of the first pixel corresponding to it is not 0, it means that there is a deviation between the formation results of the first target image and the second target image. If the error between the average values is within the set range, it is considered that the gray scale can be directly used as the final gray scale of the second pixel point; if the error exceeds the set range, it is considered that the first target image is a processing error It cannot be used as a reference. At this time, a second average value needs to be calculated, and the second pixel point is still repaired by the third pixel point around the second pixel point whose gray scale is not 0.

It is found through experiments that an error of 0.05 is a more appropriate value.

When solving the first average value, in order to obtain a more accurate result, the calculation method is set to a square area with the largest area in a circle with a second pixel as the center and rl as the radius. Preferably, rl is 2.5 times the width of a third pixel.

When calculating the second average value, r2 needs to be set to rl plus the width of a third pixel, so as to obtain a more accurate result.

When the grayscale of a first pixel is 0, calculate the third average of the grayscales of all the third pixels in the square area with the largest area in a circle with a second pixel as the center and r3 as the radius value, wherein, in the square area, the number of the third pixel points accounts for 95% of the total number of pixels, and the third average value is used as the final gray scale of the second pixel point, wherein r3 is Integer multiples of the gate.

If the gray levels of the first pixel and the second pixel are both 0, the results of the two are consistent. At this time, the first target image cannot be used as the basis for the second target image, but can only be based on the second target image itself. At this time, it is required that the number of third pixel points in the square area accounts for 95% of the total, so as to ensure that there are enough third pixel points in the square area for calculation. r3 needs to be an integer multiple of fl, preferably 2~5 times.

Step 7: After reducing all the acquired gray-scale second target images by a factor of n, they are used as right-eye images, and the initial 2D images are used as left-eye images, and the right-eye images and left-eye images are output to the 3D display device.

Although the embodiment of the present invention has been disclosed as above, it is not limited to the use listed in the specification and implementation, it can be applied to various fields suitable for the present invention, and it can be easily understood by those skilled in the art Therefore, the invention is not limited to the specific details and examples shown and described herein without departing from the general concept defined by the claims and their equivalents.

Claims (4)

1. a kind of image processing method that 2D turns 3D is characterized in that, comprises the following steps: Step 1, generating an initial depth map according to the initial 2D image; Step 2: Enlarging the initial 2D image by n times to obtain an intermediate image, and enlarging the initial depth image by n times to obtain an intermediate depth image; Step 3, using the initial 2D image and the initial depth map to obtain the first target image, and using the intermediate image and the intermediate depth map to obtain the second target image; Step 4: Obtain the coordinates of a second pixel points whose gray scale is 0 in the second target image, and the pixels whose gray scale is not 0 in the second target image are defined as third pixel points; Step 5. Calculate the coordinates of the first pixel points corresponding to a second pixel points in the first target image. The specific calculation method is: divide the coordinates of a second pixel point by n to obtain a second pixel point The coordinates of a corresponding first pixel point, so as to obtain all the first pixel points corresponding to a second pixel points one-to-one; Step 6. Obtain the gray scale of all the second pixels. The method for obtaining the gray scale of any second pixel is: When the grayscale of a first pixel is not 0, assign the grayscale of the first pixel to a corresponding second pixel, Calculate the first average value of the gray levels of all the third pixels in the circle with the second pixel as the center and rl as the radius in the square area with the largest area, wherein, in the square area, the third pixel The number of points accounts for 90% of the total number of pixels, Comparing the gray scale of a second pixel with the first average value, when the ratio of the difference between the gray scale and the first average value to the first average value is less than or equal to 0.05, the gray scale is used as the final value of the second pixel point Gray scale, when the ratio of the difference between the gray scale and the first average value to the first average value is greater than or equal to 0.05, calculate all the squares in the circle with the second pixel as the center and r2 as the radius and the largest area The second average value of the gray scale of the third pixel point, wherein the difference between r2 and rl is the width of a third pixel point, and the second average value is used as the final gray scale point of a second pixel point, When the grayscale of a first pixel is 0, calculate the third average of the grayscales of all the third pixels in the square area with the largest area in a circle with a second pixel as the center and r3 as the radius value, wherein, in the square area, the number of the third pixel points accounts for 95% of the total number of pixels, and the third average value is used as the final gray scale of the second pixel point, wherein r3 is Integer multiple of r1; Step 7: After reducing all the acquired gray-scale second target images by a factor of n, they are used as right-eye images, and the initial 2D images are used as left-eye images, and the right-eye images and left-eye images are output to the 3D display device. 2. The image processing method for converting 2D to 3D according to claim 1, wherein n is 4. 3. The image processing method for converting 2D to 3D according to claim 1, wherein r1 is 2.5 times the width of a third pixel. 4. The image processing method for converting 2D to 3D according to claim 1, wherein f3 is 2 to 5 times that of r1.

Images