Method for image shooting, terminal device, and storage medium
AssigneeGuangdong Oppo Mobile Telecommunications Corp., Ltd.
InventorYan Chen
Life of this patent
Estimating the term…
Abstract
Embodiments of the present disclosure relate to the field of electronics and provide a method for image shooting, an apparatus, a terminal device, and a storage medium. The method includes: shooting, via an image shooting component, a preview image and acquiring an exposure parameter of the preview image, when a terminal device is in a state of being ready for image shooting; obtaining an image shooting parameter for a present night scene via an image shooting parameter estimation model based on the preview image and the exposure parameter; performing, in response to receiving an image shooting instruction, image shoot processing based on the estimated image shooting parameter to obtain a target synthesized image.
Claims17
A method of image shooting, performed by a terminal device, comprising:
- shooting preview images via an image shooting component and acquiring values of an exposure parameter corresponding to the preview images, when the terminal device is in a pending state for image shooting;
- estimating a value of an image-shooting parameter of a present night scene via an estimation model according to the preview images and the values of the exposure parameter corresponding to the preview images, wherein
- the estimation model is trained by taking an image data parameter, the exposure parameter, and the image shooting parameter as variables;
- and
- the image shooting parameter comprises an image-synthesizing number, which is a number of preview images required to obtain a target synthesized-image;
- and
- performing an image-shooting process according to the estimated value of the image-shooting parameter, in response to an image-shooting instruction, to obtain the target synthesized-image;
- and
- before estimating the value of the image-shooting parameter of the present night scene via the estimation model according to the preview images and the values of the exposure parameter corresponding to the preview images, training the estimation model based on a pre-stored correspondence relationship among training preview images, values of a training exposure parameter, and values of a training image-shooting parameter in a training set, and by following a pre-stored training principle of values of the image-shooting parameter estimated by the estimation model, being approached towards values of the training image-shooting parameter corresponding to the training preview images and the values of the training exposure parameter, to achieve the estimation model.
A terminal device, comprising a processor and a non-transitory memory, wherein the non-transitory memory is arranged to store at least one instruction, at least one program, a code set, or an instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set are loaded and executed by the processor to perform operations of:
- shooting, via an image shooting component, preview images and acquiring values of an exposure parameter corresponding to the preview images, when the terminal device is in a pending state for image shooting;
- estimating, via an estimation model, a value of an image-shooting parameter of a present night scene according to the preview images and the values of the exposure parameter corresponding to the preview images, wherein
- the estimation model is trained by taking an image data parameter, the exposure parameter, and the image shooting parameter as variables;
- and
- the image shooting parameter comprises an image-synthesizing number, which is a number of preview images required to obtain a target synthesized-image;
- performing an image-shooting process according to the estimated value of the image-shooting parameter, in response to an image-shooting instruction, to obtain the target synthesized-image;
- and
- before estimating, via an estimation model, a value of an image-shooting parameter of a present night scene according to the preview images and the values of the exposure parameter corresponding to the preview images, training the estimation model based on a pre-stored correspondence relationship among training preview images, values of a training exposure parameter, and values of a training image-shooting parameter in a training set, and by following a pre-stored training principle of values of the image-shooting parameter estimated by the estimation model, being approached towards values of the training image-shooting parameter corresponding to the training preview images and the values of the training exposure parameter, to achieve the estimation model.
A non-transitory computer-readable storage medium, comprising at least one instruction, at least one program, a code set, or an instruction set stored in, wherein the at least one instruction, the at least one program, the code set, or the instruction set are capable of being loaded and executed by a processor to perform operations of:
- shooting preview images and acquiring values of an exposure parameter corresponding to the preview images;
- estimating a value of an image-shooting parameter of a present night scene via an estimation model based on the preview images and the values of the exposure parameter corresponding to the preview images, wherein the estimation model is trained by taking an image data parameter, the exposure parameter, and the image shooting parameter as variables, the image shooting parameter comprises an image-synthesizing number, which is a number of preview images required to obtain a target synthesized-image;
- performing an image-shooting process according to the estimated value of the image-shooting parameter, in response to an image-shooting instruction, to obtain the target synthesized-image;
- and
- before estimating a value of an image-shooting parameter of a present night scene via an estimation model based on the preview images and the values of the exposure parameter corresponding to the preview images, training the estimation model based on a pre-stored correspondence relationship among training preview images, values of a training exposure parameter, and values of a training image-shooting parameter in a training set, and by following a pre-stored training principle of values of the image-shooting parameter estimated by the estimation model, being approached towards values of the training image-shooting parameter corresponding to the training preview images and the values of the training exposure parameter, to achieve the estimation model.
Dependencies are inferred from the claim text.