Back to results
US 12,306,583 B2Granted
Holographic display method and holographic display system
AssigneeTcl China Star Optoelectronics Technology Co., Ltd.
InventorJinfeng Liu
Life of this patent
Estimating the term…
Abstract
A holographic display method and a holographic display system are provided. The holographic display method includes: obtaining target scene information; recognizing human face area information of the target scene information; confirming a pupil position of the human face area information; calculating a lateral viewing angle between each pixel area and each pupil position in a display panel and a driving voltage value of each pixel area at the lateral viewing angle corresponding to each pixel area; and applying the driving voltage value to the corresponding pixel area.
Claims16
A holographic display method, comprising following steps:
- S10, obtaining target scene information;
- S20, recognizing human face area information of the target scene information;
- S30, confirming a pupil position of the human face area information;
- S40, calculating a lateral viewing angle between each pixel area and each pupil position in a display panel and a driving voltage value of each pixel area at the lateral viewing angle corresponding to each pixel area;
- and
- S50, applying the driving voltage value to the corresponding pixel area;
- wherein the S20 further comprises following steps:
- S201, training a human face area detecting module;
- and
- S202, recognizing the human face area information of the target scene information by the human face area detecting module;
- wherein the S201 further comprises:
- S2011, providing an original image data set including a human face image to a feature extraction network, wherein the feature extraction network outputs original image feature information;
- and
- S2022, respectively inputting the original image feature information into an area extraction network and a classification and position feedback network, wherein the area extraction network is configured to extract a histogram of oriented gradient (HOG) feature of a predetermined area, and the classification and position feedback network is configured to classify the HOG feature and output the human face area information;
- wherein in the S2022, in the area extraction network, using a plurality of sliding windows with different sizes to encircle parts of an area having the original image feature information as a selected area and creating a region of interests (ROIs), wherein the ROIs is configured to extract the HOG feature of the selected area and output the HOG feature to the classification and position feedback network;
- wherein the step of creating the ROIs further comprises following steps:
- grouping each sub-pixel of parts of the selected area of the area having the original image feature information;
- calculating a texture feature of each group;
- and
- combining the texture feature with two groups of the sub-pixel which are the most similar to each other, wherein the ROIs is finally obtained after all sub-pixels in a certain space are combined with each other.
17 more elements
A holographic display system, comprising a light beam controlling system, a light guiding system, a display device, and a target scene obtaining system;
- wherein the target scene obtaining system is configured to obtain target scene information, the holographic display system is configured to recognize a pupil position of the target scene information, calculate a lateral viewing angle between each pixel area and each pupil position in a display panel and a driving voltage value of each pixel area at the lateral viewing angle corresponding to each pixel area, and apply the driving voltage value to the corresponding pixel area;
- wherein the holographic display system is further configured to recognize human face area information of the target scene information by training a human face area detecting module;
- wherein training the human face area detecting module comprises steps of:
- providing an original image data set including a human face image to a feature extraction network, wherein the feature extraction network outputs original image feature information;
- and
- respectively inputting the original image feature information into an area extraction network and a classification and position feedback network, wherein the area extraction network is configured to extract a histogram of oriented gradient (HOG) feature of a predetermined area, and the classification and position feedback network is configured to classify the HOG feature and output the human face area information;
- wherein in the area extraction network, steps of using a plurality of sliding windows with different sizes to encircle parts of an area having the original image feature information as a selected area and creating a region of interests (ROIs) is performed, wherein the ROIs is configured to extract the HOG feature of the selected area and output the HOG feature to the classification and position feedback network;
- wherein the step of creating the ROIs further comprises following steps:
- grouping each sub-pixel of parts of the selected area of the area having the original image feature information;
- calculating a texture feature of each group;
- and
- combining the texture feature with two groups of the sub-pixel which are the most similar to each other, wherein the ROIs is finally obtained after all sub-pixels in a certain space are combined with each other.
9 more elements
Dependencies are inferred from the claim text.