Image processing device and image processing method, display device equipped with artificial intelligence function, and method for generating a trained neural network model

Abstract

This invention provides an information processing device that utilizes artificial intelligence functions to achieve partial drive control and upward thrust control of a display device. [Solution] An information processing device that controls the operation of a display device using artificial intelligence functions comprises an acquisition unit that acquires sensor information, and an estimation unit that estimates the light emission state of each unit into which the display area of ​​the display device is divided into multiple parts, based on the sensor information, using artificial intelligence functions. The estimation unit further estimates, using artificial intelligence functions, the amount of power that is suppressed in the first unit, which is the dark part of the display area, to be distributed to the second unit, which is the bright part.

Claims

1. A trained neural network model estimates partial driving patterns that represent the individual light emission states of light source units corresponding to each of the multiple regions obtained by dividing the display area of ​​the image display unit for the target display image, A control unit that individually controls the light emission state of each light source unit based on the partial drive pattern estimated by the trained neural network model, It is equipped with, The neural network model is trained based on the error between the screen intensity distribution of the target display image input to the neural network model and the screen intensity distribution calculated based on optical propagation calculations from the partial drive pattern estimated by the neural network model. Image processing device.

2. The image display unit is a liquid crystal display unit, and during training, the display image is predicted considering the liquid crystal transmittance of the liquid crystal display unit when the light emission state of each light source unit is individually controlled based on the partial driving pattern estimated by the neural network model. The image processing apparatus according to claim 1.

3. The trained neural network model is further trained to estimate a partial drive pattern, taking into account a push-up process that distributes power suppressed in the first unit, which is the dark area of ​​the display region, to the second unit, which is the bright area. The image processing apparatus according to claim 1.

4. The trained neural network model is trained to estimate partial driving patterns for the target display image and the second information displayed on the image display unit. The image processing apparatus according to claim 1.

5. The second information includes an audio signal synchronized with the target display image. The image processing apparatus according to claim 4.

6. The second information includes at least one of the following: information obtained when decoding the video signal of the target display image or information obtained when decoding the audio signal synchronized with the video signal. The image processing apparatus according to claim 5.

7. The second information includes information about the content output by the image display unit, The image processing apparatus according to claim 4.

8. The second information includes information relating to the characteristics of the image display unit, The image processing apparatus according to claim 4.

9. The information relating to the characteristics of the image display unit includes at least one of the following: uplift conditions, viewing angle characteristics, response characteristics, or temperature characteristics. The image processing apparatus according to claim 8.

10. The second information includes information regarding the viewing environment of the image display unit, The image processing apparatus according to claim 4.

11. The second information includes information about the user viewing the image display unit. The image processing apparatus according to claim 4.

12. The second information includes information relating to the operation of the image display unit, The image processing apparatus according to claim 4.

13. The steps include: estimating a partial driving pattern for the target display image to be displayed on the image display unit using a trained neural network model that estimates partial driving patterns representing the individual light emission states of light source units corresponding to each of the multiple regions obtained by dividing the display area of ​​the image display unit with respect to the target display image; A step of controlling the light emission state of the light source unit based on the estimated partial drive pattern, It has, The neural network model is trained based on the error between the screen intensity distribution of the target display image input to the neural network model and the screen intensity distribution calculated based on optical propagation calculations from the partial drive pattern estimated by the neural network model. Image processing methods.

14. An image display unit capable of controlling the emission state of individual light source units corresponding to each of the multiple region units that divide the display area, A trained neural network model that estimates partial drive patterns representing the individual light emission states of the light source unit with respect to the target display image displayed by the image display unit, A control unit that individually controls the light emission state of each light source unit based on the partial drive pattern estimated by the trained neural network model, It is equipped with, The neural network model is trained based on the error between the screen intensity distribution of the target display image input to the neural network model and the screen intensity distribution calculated based on optical propagation calculations from the partial drive pattern estimated by the neural network model. Display device equipped with artificial intelligence functions.

15. A method for generating a trained neural network model that estimates partial driving patterns representing the light emission state of individual light source units corresponding to each of multiple regions obtained by dividing the display area of ​​an image display unit with respect to a target display image, An input step of inputting the screen intensity distribution of the target display image into a neural network model, An evaluation step which calculates the error between the screen intensity distribution of the target display image input to the neural network model and the screen intensity distribution calculated based on optical propagation calculations from the partial drive pattern estimated by the neural network model, A learning step in which the neural network model is trained based on the aforementioned error, A method for generating a trained neural network model having [a certain characteristic].

16. A first screen intensity distribution is calculated from the target display image, the contribution of each light source unit is determined by weighting the point image distribution function, which represents the intensity distribution formed by each light source unit of the image display unit, with the predicted value of the partial driving pattern estimated by the neural network model from the first screen intensity distribution, the contribution of each light source unit is calculated by summing the contributions obtained from all light source units, and a second screen intensity distribution is calculated when each light source unit of the image display unit is driven, and a loss function defined based on the error between the first screen intensity distribution and the second screen intensity distribution is calculated. In the learning step, the neural network model is trained using the loss function. A method for generating a trained neural network model according to claim 15.

17. The optical propagation calculation uses a plurality of point image distribution functions corresponding to the physical positions of the light source units, and calculates the screen intensity distribution by summing the intensity contributions from the light source units based on the plurality of point image distribution functions. The image processing apparatus according to claim 1.

Images