阅读说明：本技术 一种基于多层透明液晶屏的裸眼3d显示方法及装置 (Naked eye 3D display method and device based on multilayer transparent liquid crystal screen ) 是由 凌涛 凌婧 于 2020-07-15 设计创作，主要内容包括：一种基于多层透明液晶屏的裸眼3D显示方法及装置,装置包括多层透明液晶屏,方法包括单目图像深度图估计,以及深度向层次转换两个部分,单目图像深度图估计通过对部分视频帧的深度信息计算,关联计算得到一段具有深度信息的视频帧,深度向层次转换依据深度信息结合考虑多层液晶屏间距离和屏幕间的光线衰减,把图像中不同深度的物体内容对应分配到多层透明液晶屏的各个液晶屏上进行显示,实现裸眼3D显示。本发明基于经典的SFS算法原理,对其计算过程进行了优化,提高了其计算速度,依据液晶多层屏的特点,裁剪了计算出来的深度信息,达到了基本能够实时计算图像内容的深度计算,满足了工程化应用的要求。(A naked eye 3D display method and device based on a multilayer transparent liquid crystal screen comprise the multilayer transparent liquid crystal screen, the method comprises monocular image depth map estimation and depth-to-level conversion, the monocular image depth map estimation is used for calculating the depth information of partial video frames, a section of video frame with depth information is obtained through correlation calculation, the depth-to-level conversion is carried out according to the depth information and by considering the distance between the multilayer liquid crystal screens and the light attenuation between the screens, object contents with different depths in the image are correspondingly distributed to each liquid crystal screen of the multilayer transparent liquid crystal screen to be displayed, and naked eye 3D display is achieved. Based on the classic SFS algorithm principle, the calculation process is optimized, the calculation speed is increased, the calculated depth information is cut according to the characteristics of the liquid crystal multi-layer screen, the depth calculation capable of basically calculating the image content in real time is achieved, and the requirements of engineering application are met.)

1. A naked eye 3D display method based on a multilayer transparent liquid crystal screen is characterized by comprising a monocular image depth map estimation part and a depth-to-level conversion part,

monocular image depth map estimation, a plurality of frame pictures are extracted at intervals in a continuous video frame Img1, the extracted pictures are divided into a plurality of regions by a level set method to obtain depth information of each region, the pictures with the region depth information are used as stable frames Img3, other video frames Img1 ' in the video frame Img1 are also divided into regions by the level set method, the region of Img1 ' is corresponding to the region of Img3 according to space-time relevance, the region depth information of Img3 is assigned to the corresponding region of Img1 ' to obtain a picture set Img2 with the depth information, and Img3 with the depth information and the assigned Img2 are combined according to the video frame sequence to obtain a continuous video frame DImg1 with the depth information as output;

and (4) converting the depth into the hierarchy, and correspondingly distributing object contents of different depths in the image to each liquid crystal screen for displaying by considering the distance between the multiple layers of liquid crystal screens and the light attenuation between the screens according to the DImg1 with the depth information.

2. The naked eye 3D display method based on the multilayer transparent liquid crystal display screen as claimed in claim 1, wherein the monocular image depth map estimation specifically comprises:

step 1-1, numbering video frame images according to time, wherein Img1 (i) represents the ith frame of video frame image;

step 1-2, extracting a plurality of frames of pictures Img3(j) from Img1 (i) according to a set interval, wherein j belongs to i, each extracted picture is divided into regions by adopting a level set method, each pixel is divided into a plurality of different regions Ori (j, f), Ori (j, f) represents the f-th region in the video frame Img3(j), the central position region is Cen-Ori (j) = (wpx, hpy), wpx and hpy represent two position parameters of the width and height of the central position region, and when the level set division is adopted, the proportion of each region in the whole image area is not lower than a set value T1%;

step 1-3, uniformly sampling pixels in each area Ori (j, f) to obtain a set Pc (j, f) = { P (j, f, P) } of N sampling points in the area, wherein P represents the number of the sampling point P, P = 1-N, and Pc is the set of the sampling points, and calculating Depth information Depth1(j, f, P) of the sampling points by adopting a method of a light and shade recovery shape SFS;

step 1-4, averaging Depth1(j, f, p) information of sampling points to obtain an average value Depth2 (j, f), taking Depth2 (j, f) as the Depth value of the f-th area in the video frame Img3(j), and sequentially calculating the Depth values of all areas to obtain Depth information Img3(j) with an integral image;

step 1-5, performing regionalization division on the video frame Img1 '(k) left after extraction by a level set method, wherein k belongs to i, k is not equal to j, and the region is represented as Ori (k, g), namely the g-th region in the video frame Img 1' (k), then corresponding the region of Img1 'to the region of Img3 based on temporal-spatial correlation, and reducing the calculation amount of the depth information of Img 1' by an assignment method; the temporal spatial correlation refers to: the time interval between Img 1' and Img3 does not exceed a set value T2, and the spatial distance of the central area of the area does not exceed a set value T3, specifically: for video frame Img1 ' (k) at time interval Ng < T2 from Img3(j), if the center position area Cen-ori (k) = (wpx _ k, hpy _ k) of the divided areas and the center position area Cen-ori (j) = (wpx, hpy) of Img3(j) have a positional deviation and | wpx-wpx _ k | + | hpy-hpy _ k | less than T3, Img1 ' (k) is associated with Img3(j), the area depth information of the associated Img3(j) is assigned to Img1 ' (k), resulting in video frame Img2(k) with the assigned depth information;

and 1-6, arranging the video frames Img3(j) and Img2(k) according to the video frame sequence to obtain continuous video frames DImg1 and output.

3. The naked eye 3D display method based on the multilayer transparent liquid crystal display screen as claimed in claim 2, wherein monocular image depth map estimation is performed by taking 15 frames of video frames as a group, and depth information calculation is performed by extracting the video frames at intervals of 5 frames.

4. The naked eye 3D display method based on the multilayer transparent liquid crystal display screen as claimed in claim 1 or 2, wherein the depth-to-level conversion is specifically as follows: the liquid crystal multi-layer screen is provided with 1 bottom layer common liquid crystal screen and at least 1 transparent liquid crystal screen, and the total number of the screens is X;

step 2-1, if X =2, namely only 1 bottom layer common liquid crystal screen and 1 layer transparent liquid crystal screen, directly classifying by the intermediate value of the depth information of each region in the video frame image, taking the region larger than the intermediate value as the background, and taking the region smaller than or equal to the intermediate value as the foreground, wherein the bottom layer common liquid crystal screen displays the background in the video frame, and the transparent liquid crystal screen displays the foreground;

step 2-2, if X is greater than 2, according to the depth reverse log calculation, setting the display content of each transparent liquid crystal screen, namely distributing the area of each video frame image in the monocular image depth map estimation on the transparent liquid crystal screen for display according to the depth information, specifically:

step 2-2-1, let DImg1 (i) beiIn a video frame with Depth information, the Depth of each region is Depth2 (i, f), and the minimum value DepthF (i) of the depths of all the regions is obtained;

step 2-2-2, the Depth2 (i, f) of each object content in the image is logarithmically processed to obtain a new Depth expression Depth3(i, f), namely:

obtaining new Depth information Depth3(i, f) of each object;

step 2-2-3, averagely dividing Depth3(i, f) into X sections, calculating which section the Depth3(i, f) value of each area Ori (i, f) belongs to, and then distributing the content of the Ori (i, f) object to the transparent liquid crystal screen of the corresponding layer;

and 2-2-4, filling the non-bottom liquid crystal screen to obtain the object content required to be displayed in each frame of each liquid crystal screen, and filling the rest non-object content in each frame of image into white with minimum brightness in the display of the non-bottom liquid crystal screen.

5. The naked eye 3D display method based on the multilayer transparent liquid crystal screen according to claim 1, wherein computing resources of monocular image depth map estimation and depth to hierarchy conversion are divided, computing resources of monocular image depth map estimation are provided through a cloud computer in combination with a cloud technology, computing results of monocular image depth map estimation are transmitted to a hierarchical computing processing module carried by a local liquid crystal screen, and the hierarchical computing processing module only needs to perform hierarchical computing according to received depth information and then drive the liquid crystal screen to display, so that a naked eye 3D display effect can be obtained.

6. A naked eye 3D display device based on a multilayer transparent liquid crystal screen is characterized by comprising 1 bottom layer liquid crystal screen, at least 1 transparent liquid crystal screen, a layering calculation module and a liquid crystal driver, wherein the bottom layer liquid crystal screen is used for displaying the background part and the opacity of each frame of picture in a video, the transparent liquid crystal screen is used for displaying the contents of different foreground layers in each frame of picture in the video, the bottom layer liquid crystal screen and the transparent liquid crystal screen are both provided with the contents to be displayed by the liquid crystal driver, the layering calculation module is in bidirectional data connection with the liquid crystal driver, for providing computing resources, a computer program is configured in the hierarchical computing module, and when the computer program is executed, the naked eye 3D display method according to any one of claims 1-5 is realized, after the video data is acquired, and outputting the distribution scheme of the layered contents to a liquid crystal driver, and driving the corresponding common liquid crystal screen and the transparent liquid crystal screen to display.

7. The naked eye 3D display device based on the multilayer transparent liquid crystal screen as claimed in claim 6, wherein the distance between the bottom layer liquid crystal screen and the transparent liquid crystal screen of the naked eye 3D display device is equal to 100mm in distance from front to back.

8. The naked-eye 3D display device based on the multilayer transparent liquid crystal screen is characterized in that the display device is connected with a cloud server, the cloud server provides calculation resources for monocular image depth map estimation, calculation results of the monocular image depth map estimation are transmitted to a layered calculation module of the display device, the layered calculation module outputs a distribution scheme of layered contents to a liquid crystal driver according to the received calculation results, and the corresponding common liquid crystal screen and the corresponding transparent liquid crystal screen are driven to display.

Technical Field

The invention belongs to the technical field of liquid crystal screens, relates to naked eye 3D display, and discloses a naked eye 3D display method and device based on a multilayer transparent liquid crystal screen.

Background

Currently, the mainstream naked eye 3D technical means includes: slit type liquid crystal grating, columnar lens, directional light source and active backlight.

[1] Slit type liquid crystal grating. The technical principle is that after a slit type grating is added in front of a screen, when an image which is to be seen by a left eye is displayed on a liquid crystal screen, an opaque stripe can shield the right eye, and similarly, when the image which is to be seen by the right eye is displayed on the liquid crystal screen, the opaque stripe can shield the left eye, and a viewer can see a 3D image by separating visual pictures of the left eye and the right eye.

[2] The technical principle of the cylindrical lens is that pixel points corresponding to the left eye and the right eye are projected into the left eye and the right eye respectively through the refraction principle of the lens, and image separation is achieved. The biggest advantage of the contrast slit grating technology is that the lens does not block light, so the brightness is greatly improved.

[3] And directing to the light source, namely precisely controlling the two groups of screens to respectively project images to left and right eyes.

[4] The active backlight module is formed by adopting an optical microstructure, and the direction of light beams emitted by the backlight module can be adjusted under the control of an electronic device.

These are all naked eye 3D technologies constructed based on optical methods. Nowadays, transparent screen technology is mature continuously, and the cost is lower and lower. If the display equipment consists of a plurality of layers of transparent screens and a layer of bottom layer common liquid crystal screen, the contents of different distances in a scene can be embodied on different layers, and normal naked-eye 3D display can be directly realized in principle.

Disclosure of Invention

The technical problem to be solved by the invention is as follows: the liquid crystal multi-layer screen can be used for realizing naked eye 3D display through depth of field, but in the prior art, no proper solution is provided for how to divide image depth in real time under the self-contained computing capability of the liquid crystal screen, so that the liquid crystal multi-layer screen displays a 3D image.

The technical scheme of the invention is as follows: a naked eye 3D display method based on a multilayer transparent liquid crystal screen comprises monocular image depth map estimation and depth-to-hierarchy conversion,

monocular image depth map estimation, a plurality of frame pictures are extracted at intervals in a continuous video frame Img1, the extracted pictures are divided into a plurality of regions by a level set method to obtain depth information of each region, the pictures with the region depth information are used as stable frames Img3, other video frames Img1 ' in the video frame Img1 are also divided into regions by the level set method, the region of Img1 ' is corresponding to the region of Img3 according to space-time relevance, the region depth information of Img3 is assigned to the corresponding region of Img1 ' to obtain a picture set Img2 with the depth information, and Img3 with the depth information and the assigned Img2 are combined according to the video frame sequence to obtain a continuous video frame DImg1 with the depth information as output;

and (4) converting the depth into the hierarchy, and correspondingly distributing object contents of different depths in the image to each liquid crystal screen for displaying by considering the distance between the multiple layers of liquid crystal screens and the light attenuation between the screens according to the DImg1 with the depth information.

The invention also provides a naked eye 3D display device based on the multilayer transparent liquid crystal screen, which comprises 1 bottom layer liquid crystal screen, at least 1 transparent liquid crystal screen, a layering calculation module and a liquid crystal drive, wherein the bottom layer liquid crystal screen is used for displaying the background part and the opacity of each frame of picture in the video, the transparent liquid crystal screen is used for displaying the contents of different foreground layers in each frame of picture in the video, the bottom layer liquid crystal screen and the transparent liquid crystal screen are both provided with the contents to be displayed by the liquid crystal drive, the layering calculation module is in bidirectional data connection with the liquid crystal drive and is used for providing calculation resources, and a computer program is configured in the layering calculation module.

Based on the classic SFS algorithm principle, the calculation process is optimized, the calculation speed is increased, the calculated depth information is cut according to the characteristics of the liquid crystal multi-layer screen, the depth calculation capable of basically calculating the image content in real time is achieved, and the requirements of engineering application are met.

Drawings

Fig. 1 is a flowchart of video monocular image depth map estimation according to an embodiment of the present invention.

Fig. 2 is a flowchart of depth-to-hierarchy conversion according to an embodiment of the present invention.

Fig. 3 is a schematic diagram of the relationship among the contents of video, frame and object according to the embodiment of the present invention.

Fig. 4 is a schematic diagram of a hardware structure according to an embodiment of the present invention.

Detailed Description

The invention provides a display method and a device thereof capable of realizing naked eye 3D in principle based on the existing artificial intelligence and electronic technology. The multi-layer liquid crystal display screen consists of a plurality of layers of transparent screens and a layer of bottom layer common liquid crystal screen, and can embody the contents of different distances in a scene on different layers; the depth of field of various targets in each frame of video picture can be automatically calculated and distributed to different transparent liquid crystal screens and bottom liquid crystal screens for displaying, and normal naked-eye 3D display is realized in principle.

The scheme of the invention is realized based on the liquid crystal multi-layer screen, and the display method is explained as follows.

The display method comprises monocular image depth map estimation and depth-to-hierarchy conversion.

1. Monocular image depth map estimation

The monocular image depth map estimation algorithm is used for estimating the depth of contents in a scene by utilizing a two-dimensional image, is one of classic problems in the field of computer vision, and is also an important link for realizing three-dimensional reconstruction and scene perception. The invention aims to realize the effect of calculating depth information in real time based on the calculation capability of the existing transparent liquid crystal screen and meet the engineering requirement, so that the main means is to cut the light and shade recovery shape SFS in the existing traditional method.

As the classical SFS calculates the depth information of the object surface by utilizing the brightness and the shadow of the image under different illumination conditions for each frame and each pixel of the video, the calculation amount is extremely large, the result is fine, but for the hardware basis of the liquid crystal multi-layer screen, the fine calculation exceeds the accuracy range which can be reached by the hardware, so the invention achieves the purpose of reducing the calculation amount by abandoning the accuracy reached by the classical SFS, thereby achieving the engineering goal of real-time calculation.

The invention extracts some pictures at intervals in a video frame, divides the image into a plurality of regions by a level set method, obtains the depth information of each region by the calculation of a cut SFS algorithm, and takes the pictures with the region depth information as relatively accurate frames, namely stable frames. For other video frame pictures, in a region with small time interval with the stable frame and small space distance interval with the central position region, a picture set with assigned depth information is obtained by directly adopting a depth information endowing mode of the stable frame, and a group of video frame picture sets with depth information of each frame are obtained by combining the stable frame and other video frames after assignment with a small amount of calculation and are used as output.

The process is shown in fig. 1, wherein the SFS algorithm calculation process is a classical algorithm, which is not described in detail herein.

Step 1-1, numbering the video frame images according to time, wherein Img1 (i) represents the ith frame of video frame picture.

Step 1-2, extracting a plurality of frame pictures Img3(j) in Img1 (i) according to a set interval, wherein j belongs to i, preferably 15 frames of video frames are taken as a group to carry out monocular image depth map estimation, and extracting the video frames at an interval of 5 frames to carry out depth information calculation. For each extracted picture, firstly, a level set method is adopted to perform regional division on the whole image, each pixel is divided into a plurality of different regions Ori (j, f), the Ori (j, f) represents the f-th region in the video frame Img3(j), the central position region is Cen-Ori (j) = (wpx, hpy), wpx and hpy represent two position parameters of the width and the height of the central position region, when the level set division is adopted, the proportion of each region in the whole image area is not lower than a set value T1%, and in the embodiment, T1 is 5.

Step 1-3, uniformly sampling pixels in each area Ori (j, f) to obtain a set of N sampling points Pc (j, f) = { P (j, f, P) }, wherein P represents the number of the sampling point P, P = 1-N, and Pc is the set of the sampling points, and calculating the Depth information Depth1(j, f, P) of the sampling points by adopting a method of a light and shade recovery shape SFS.

Step 1-4, averaging Depth1(j, f, p) information of sampling points to obtain an average value Depth2 (j, f), wherein the average value has the effects of filtering and removing noise, Depth2 (j, f) is used as the Depth value of the f-th area in the video frame Img3(j), and the Depth values of all the areas are calculated in sequence to obtain Depth information Img3(j) with the whole image.

Step 1-5, performing regionalization division on the video frame Img1 ' (k) left after extraction by a level set method, wherein the region of k ∈ i, k ≠ j, j ≠ k = i, and Img1 ' (k) is represented as Ori (k, g), namely the g-th region in the video frame Img1 ' (k), then corresponding the region of Img1 ' to the region of Img3 based on time-space relevance, and reducing the calculation amount of the depth information of Img1 ' in an assignment manner; the temporal spatial correlation refers to: the time interval between Img 1' and Img3 does not exceed a set value T2, and the spatial distance of the central area of the area does not exceed a set value T3, specifically: for video frame Img1 ' (k) at time interval Ng < T2 of Img3(j), if the central position area Cen-ori (k) = (wpx _ k, hpy _ k) of the divided areas and the central position area Cen-ori (j) = (wpx, hpy) of the divided areas and | wpx-wpx _ k | + | hpy-hpy _ k | are smaller than T3, this Img1 ' (k) is associated with Img3(j), and for consecutive video frames, in the case of two frames of spatio-temporal association, the area division may be regarded as the same, the area depth information of the associated Img3(j) is assigned to Img1 ' (k), resulting in video frame Img2(k) with assigned depth information.

In this embodiment, T2 is preferably 5, i.e. within 5 frames; t3 is 30 pixels. If the time distance and the space distance between a video frame to be assigned and two previous and next stable frames are within the range of the set value, one frame can be optionally assigned, and in the actual programming, the previous stable frame is preferably used for assignment.

And 1-6, arranging the video frames Img3(j) and Img2(k) according to the video frame sequence to obtain continuous video frames DImg1 and output.

Examples are as follows:

inputting: for a set of 15 original video frame pictures, frame number i is {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15}, which contains no depth information. Drawing at intervals of 5, wherein the frame number j is {1, 6 and 11 }; calculating a segmentation area and corresponding depth information of a video frame with frame numbers {1, 6 and 11} to obtain a stable frame Img 3; then, fitting the depth information of the rest frames k by utilizing the space-time correlation among the continuous video frames to obtain the video frame sequence with all the depth information: DImg1= {1 (depth calculation), 2 (depth assignment), 3 (depth assignment), 4 (depth assignment), 5 (depth assignment), 6 (depth calculation), 7 (depth assignment), 8 (depth assignment), 9 (depth assignment), 10 (depth assignment), 11 (depth calculation), 12 (depth assignment), 13 (depth assignment), 14 (depth assignment), 15 (depth assignment) }, compared with video frames {1, 6, 11} having higher depth information accuracy, and other video frames in which depth information is obtained by assignment have relatively lower depth information accuracy.

Compared with the classical SFS algorithm, the method carries out a large amount of calculation quantity reduction work, which is mainly embodied in that for each frame of picture, the calculation of the SFS is not carried out on each pixel; for consecutive frame pictures, the calculation of SFS is not performed per area either. Assuming that a picture is 1024 × 768, 10 regions are reserved for each picture, and 5 points are sampled for each region to calculate SFS, Ng =5 frames, only about 10 (regions) × 5 (points)/(1024 × 768 × 10) =1.2 × 10 may be reserved^-5The computing requirements on the SFS are greatly reduced, thereby greatly reducing the burden on hardware.

As shown in fig. 3, video 1 is video data Img1 input to a liquid crystal multi-layer screen, a frame of image is extracted from the video to obtain video frame 2, i.e., Img3, black in the video frame 2 is a background, triangles and ellipses have different depth information, a depth information division diagram 3 is obtained, and the image background, the triangles and the ellipses are divided into 3 layers, i.e., a background 4, a triangular layer 5 and an elliptical layer 6.

The depth-to-hierarchy conversion is a method for mapping each object content to each liquid crystal screen according to depth, considering the distance between multiple layers and the attenuation of rear light rays passing through a front screen on the basis of obtaining the depth information of each frame of image content in a video, as shown in fig. 2, specifically comprising:

the liquid crystal multi-layer screen is provided with 1 bottom layer common liquid crystal screen and at least 1 transparent liquid crystal screen, and the total number is X screens.

And 2-1, if X =2, only 1 bottom layer of common liquid crystal display and 1 transparent liquid crystal display are provided. Then directly classifying by the intermediate value of the depth information of each area in the video frame image, wherein the area larger than the intermediate value is a background, and the area smaller than or equal to the intermediate value is a foreground. The bottom layer common liquid crystal screen displays the background in the video frame, and the transparent liquid crystal screen displays the foreground.

And 2-2, if X is greater than 2, setting the display content of each transparent liquid crystal screen according to the depth inverse log calculation, namely distributing the area of each video frame image in the monocular image depth map estimation on the transparent liquid crystal screen according to the depth information of the area to be displayed, wherein the reason for setting the log is to consider the transmission and diffusion of light of the rear liquid crystal screen when passing through the front liquid crystal screen, so that the depth stratification is more natural.

Step 2-2-1, let DImg1 (i) beiIn the video frame with Depth information, the Depth of the f-th area is Depth2 (i, f), and the Depth value which is the closest distance to the observation point in all the f areas is Depth f (i).

Step 2-2-2. due to the perception of human eyes on distance transformation and the transmission and diffusion of the light of the liquid crystal screen when passing through the liquid crystal screen, the Depth Depth2 (i, f) and the total logarithm DepthF (i) of the content of each object are taken down to obtain a new Depth expression Depth3(i, f), namely

This results in new Depth information Depth3(i, f) for each object.

And 2-2-3, using the number 1 to represent the frontmost transparent liquid crystal screen, using the number to code each layer of liquid crystal screen, and using the liquid crystal screen with the background information of the number X, namely the bottommost background liquid crystal screen, equally dividing the Depth3(i, f) into X sections, and calculating which section the Depth3(i, f) value of each Ori (i, f) belongs to. The Ori (i, f) object content is then distributed to the corresponding transparent liquid crystal screen.

And 2-2-4, filling the non-bottom liquid crystal screen, wherein the object content required to be displayed in each frame of each liquid crystal screen can be obtained through the step 2-3, and for other residual non-object contents in each frame of image, white with minimum brightness is filled in the display of the non-bottom liquid crystal screen.

Depth2 (i, f) can be mapped to corresponding X segments by building a Depth segmentation table and building a table look-up method, thereby reducing the requirement of computing resources. The log function is maintained, i.e. the transmission and diffusion of the light from the rear panel through the front panel, while the log calculation is hardly increased.

Thus, new video frames with depth are distributed to X liquid crystal screens for display, and the new videos form a naked eye 3D effect through display of the liquid crystal screens, and the effect is optimal at a central point vertical to the equipment.

On the basis of the display method, the invention provides a naked eye 3D display device, as shown in FIG. 4, which comprises 1 bottom layer liquid crystal screen 7, at least 1 transparent liquid crystal screen 8, a layering calculation module 9 and a liquid crystal driver 10. The bottom liquid crystal screen is used for displaying a background part of each frame of picture in the video, the transparent liquid crystal screen is used for displaying contents of different foreground layers in each frame of picture in the video, the bottom liquid crystal screen and the transparent liquid crystal screen are both provided with contents to be displayed by a liquid crystal driver, the layering calculation module is in bidirectional data connection with the liquid crystal driver and is used for providing calculation resources, and a computer program is configured in the layering calculation module, and when the computer program is executed, the naked eye 3D display method is realized, after video data is acquired, a distribution scheme of the layering contents is output to the liquid crystal driver, and corresponding common liquid crystal screens and the transparent liquid crystal screens are driven to display.

Preferably, the distance between the bottom layer liquid crystal screen and the transparent liquid crystal screen of the naked eye 3D display device is equal to 100mm in distance from front to back.

The method of the invention compares the advantageous effects of the prior art with one specific example. In this embodiment, taking the hierarchical computer as a dell T7920 graphics workstation with a GPU as an example, the calculation cycle of the hierarchical algorithm in the prior art is very long, and a common computer with a GPU needs 5 seconds to process one frame of picture, and for a video with 1024 × 768 resolution and 30 frames per second, 1080000 seconds is needed. Whereas the computational resources of the method of the invention require only about 1.2 x 10 of it^-5The time is only 13 seconds, and with the development of computing chips, sufficient computing power can be provided even by computing resources carried by a common liquid crystal display. Further, the method comprisesThe method is divided into two parts, can be divided, and can be combined with the cloud technology in the prior art, the result is transmitted to a local general layered calculation processing module after monocular image depth map estimation calculation resources are provided by a cloud high-performance computer, the 3D effect can be obtained only by performing layered calculation according to depth information and then driving and displaying, and the scheme can also reduce the cost.