Multimedia children's picture book generation method and system based on generative artificial intelligence

Abstract

This invention discloses a method and system for generating multimedia children's picture books based on generative artificial intelligence. The method includes: acquiring a scanned image of a target physical picture book; identifying and extracting text information from the image; analyzing and extracting the subject, actions, and corresponding descriptive information; detecting the main object in the image and separating it from the background; generating a Stable-Diffusion model's introductory text based on the detected main object, the extracted subject, actions, and corresponding descriptive information, and combining it with a preset template; generating a dynamic image using a pre-trained Stable-Diffusion model; and completing the background image of the separated main object; inputting the dynamic image into a SwinBERT video summarization module to obtain a corresponding video description summary; and synthesizing corresponding narration, dialogue, or background sound based on the content of the dynamic image or pre-set content, thereby transforming a static picture book into a generative multimedia children's picture book. This method helps provide children with a personalized and interactive experience.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of image, text and voice generated artificial intelligence, and more particularly relates to a multimedia children's picture book generation method and system based on generated artificial intelligence. BACKGROUND

[0002] Currently, children's picture books play an important role in cultivating children's reading interest, inspiring imagination and developing language ability. In modern society, the demand for innovation in the field of children's education and entertainment is increasing. Although traditional paper children's picture books have affinity and tactile experience, their content and form are limited. Traditional children's picture books usually only contain static images and text, lack dynamic and sound elements, and are difficult to attract children's attention, and cannot provide personalized and interactive experience. Therefore, a system capable of generating multimedia children's picture books is needed to provide a more rich and lively reading experience.

[0003] With the development of generated artificial intelligence technology, a multimedia children's picture book generation system based on this technology has emerged. The multimedia children's picture book generation system based on generated artificial intelligence can realize an automated picture book re-creation process, providing children with personalized, interactive and interesting reading and entertainment experience.

[0004] Therefore, how to combine generated artificial intelligence technology to re-create static picture books into multimedia children's picture books and provide children with personalized, interactive and interesting reading and entertainment experience is a problem that needs to be solved by those skilled in the art. SUMMARY

[0005] Therefore, the present application provides a multimedia children's picture book generation method and system based on generated artificial intelligence, which can realize an automated re-creation process based on static picture books, providing children with personalized, interactive and interesting reading and entertainment experience.

[0006] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0007] The embodiment of the present application provides a multimedia children's picture book generation method based on generated artificial intelligence, comprising the following steps:

[0008] S1, obtaining the scanned image of the target entity picture book, recognizing and extracting the text information in the image;

[0009] S2, analyzing the sentence syntax of the text information, and extracting the subject, action and corresponding description information;

[0010] S3, detecting the subject object in the image;

[0011] S4, separating the subject object and the background in the image;

[0012] S5, generating a guide language of a Stable-Diffusion model according to the subject object detected in the image, and the subject, action and corresponding description information extracted by analyzing the text information, and combining a preset template;

[0013] S6, generating a dynamic image from a static picture of the subject object separated from the image by using a pre-trained Stable-Diffusion model, and completing a background picture of the separated subject object;

[0014] S7, inputting the dynamic image into a SwinBERT video summary module to obtain a corresponding video description summary;

[0015] S8, synthesizing corresponding narration, dialogue or background sound according to the content of the dynamic image or pre-set content, so as to convert the static picture into a generative multimedia children's picture book.

[0016] Further, the step S1 comprises:

[0017] scanning each page of a paper target entity picture book to obtain a digital image, or directly obtaining each digital image of an electronic target entity picture book;

[0018] performing OCR text detection and recognition extraction on the digital image, and recording the position of the text in the digital image.

[0019] Further, the step S2 comprises:

[0020] performing syntax analysis on the text information by using a pre-trained NLP syntax analysis model, and extracting the subject, action and corresponding description information in the description sentence in the picture book.

[0021] Further, the step S3 comprises:

[0022] performing object recognition in the image by using an object recognition model pre-trained for children's picture books;

[0023] recognizing the subject object in the image and its position in the image.

[0024] Further, the step S4 comprises:

[0025] performing semantic segmentation of the image by taking the subject object in the image as the foreground;

[0026] obtaining the mask pixels of the subject object in the image, and extracting the subject object from the background;

[0027] Add z-axis information to the two-dimensional coordinates of the main object in the image to convert it into three-dimensional coordinates; if overlapping relationships are found between objects, then distinguish them based on the z-axis information.

[0028] The main object is removed from the image to obtain a background image; the z-axis of the background image is set to reference 0.

[0029] Furthermore, step S5 also includes:

[0030] If the main object matches the subject in the syntactic analysis of a sentence, the preset template is: generate a series of images based on the subject, action, and corresponding descriptive information to complete its introductory text;

[0031] If the target object cannot match the subject in the syntactic analysis, or if no verb indicating an action is found in the syntactic analysis, the preset template is: based on the subject and the corresponding descriptive information, a preset action is set, and the guide text generated by the continuous images is completed based on the action.

[0032] Further, step S6 includes:

[0033] S61. Input the background image or the occluded foreground image separated from the image into the pre-trained Stable-Diffusion model to complete the static image.

[0034] S62. Input the extracted main object image, or the extracted and completed main object image, and the guiding text into the pre-trained Stable-Diffusion model to generate a series of multi-frame images of the corresponding object and the corresponding action.

[0035] S63. Align the multi-frame images generated from multiple groups of objects along the timeline and stitch them into the completed static background image; during stitching, use the z-axis information to maintain the overlapping relationship between the extracted objects.

[0036] S64. Generate multi-frame images of continuous motion according to the above steps S61 to S63; play the multi-frame images continuously in a loop according to the set frame rate to create a page in the dynamic picture book.

[0037] Further, step S8 includes:

[0038] Based on the obtained video description summary, the LLM model is invoked to obtain the corresponding background sound type; a matching search is performed from the background sound library to obtain the corresponding background sound;

[0039] For a children's picture book, a pre-trained speech synthesis VALL-E model of various cartoon and animal characters is trained; according to the video description abstract, if the object character in the picture has a dialogue, the matching character features are searched from the character voice library to find the most matching voice model; according to the text content extracted by the syntax analysis, the dialogue voice is synthesized;

[0040] If the text in the picture is a voiceover, a preset voice model is used to generate the voiceover;

[0041] The dialogue and the voiceover are sorted according to the time sequence, and then fused with the background sound to generate a voice file of the current page of the picture book; the voice file and the content of the dynamic image are combined to generate the final video file of the current page of the picture book.

[0042] In a second aspect, the embodiments of the present application also provide a multimedia children's picture book generation system based on generative artificial intelligence, comprising:

[0043] A scanning and extracting module is configured to obtain an image scanned by a target entity picture book, and identify and extract text information in the image;

[0044] A syntax analysis module is configured to analyze the syntax of the text information, and extract subject, action and corresponding description information;

[0045] An image object detection module is configured to detect the subject object in the image;

[0046] An image semantic segmentation module is configured to separate the subject object and the background in the image;

[0047] A guided text generation module is configured to generate a guided language of a Stable-Diffusion model according to the subject object detected in the image, and the subject, action and corresponding description information extracted by analyzing the text information, in combination with a preset template;

[0048] A dynamic image generation module is configured to generate a dynamic image from a static picture of the separated subject object in the image by using a pre-trained Stable-Diffusion model; and complete the background picture of the separated subject object;

[0049] A video abstract description module is configured to input the dynamic image into a SwinBERT video abstract module to obtain a corresponding video description abstract;

[0050] A speech synthesis and generation module is configured to synthesize corresponding voiceover, dialogue or background sound according to the content of the dynamic image or pre-set content, so as to convert the static picture book into a generative multimedia children's picture book.

[0051] As can be seen from the above technical solutions, compared with existing technologies:

[0052] Traditional printed children's picture books are limited in content and format, failing to provide a personalized and interactive experience. To address this issue, this invention utilizes generative artificial intelligence technology, combining image scanning, text recognition, syntactic analysis, image object recognition, image semantic segmentation, automatic text generation, background completion image generation, dynamic image generation, and speech synthesis to automate the generation process of multimedia children's picture books, helping to provide children with a personalized and interactive experience. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0054] Figure 1 The flowchart provides a method for generating multimedia children's picture books based on generative artificial intelligence for this invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] This invention discloses a method for generating multimedia children's picture books based on generative artificial intelligence, referring to... Figure 1 As shown, it includes the following steps:

[0057] S1. Obtain the scanned image of the target physical picture book, and identify and extract the text information in the image;

[0058] S2. Analyze the syntax of the text information to extract the subject, action, and corresponding descriptive information;

[0059] S3. Detect the main object in the image;

[0060] S4. Separate the main object and the background in the image;

[0061] S5. Based on the main object detected in the image, and the main body, action and corresponding description information extracted by analyzing the text information, combined with the preset template, generate the prompts for the Stable-Diffusion model.

[0062] S6. Using a pre-trained Stable-Diffusion model, the static image of the main object separated from the image is used to generate a dynamic image; and the background image of the separated main object is completed.

[0063] S7. Input the dynamic image into the SwinBERT video summarization module to obtain the corresponding video description summary;

[0064] S8. Based on the content of the dynamic image or the pre-set content, synthesize the corresponding narration, dialogue or background sound, thereby transforming the static picture book into a generative multimedia children's picture book.

[0065] The following sections will provide a detailed explanation of each of the above steps.

[0066] In step S1 above, the specific steps involve scanning the physical picture book in high definition; this step can be omitted for electronic picture books. Each page of the picture book is then converted into a high-definition digital image. Subsequent operations will use one of these images as an example. Then, OCR text detection and recognition are performed on the high-definition scanned picture book image or electronic picture book image, and the position of the text within the image is recorded.

[0067] In step S2 above, the specific contents include:

[0068] S21. A pre-trained NLP model (Neuro-Linguistic Programming model) is used to perform syntactic analysis on the text extracted by OCR;

[0069] S22. Syntactic analysis can obtain the subject, action, and descriptive words in the descriptive sentences of the picture book.

[0070] In step S3 above, the specific contents include:

[0071] S31. Use an object recognition model pre-trained for children's picture books to perform object recognition in the whole image; for example, use RCNN, Faster RCNN, SSD, RPN, etc., this disclosure embodiment does not limit this;

[0072] S32. Identify the main object in the image and its position in the image.

[0073] In step S4 above, the specific contents include:

[0074] S41. Use the objects identified in S3 as the foreground to perform semantic segmentation of the image.

[0075] S42. Obtain the mask pixels of these objects in the image and extract the objects from the background.

[0076] S43. Add z-axis information to the 2D coordinates of objects in the image, converting them to 3D coordinates. If the extracted objects overlap, they need to be differentiated in terms of z-axis information. If the extracted areas of two objects are connected by an edge, the z-axis values ​​of the two objects need to be distinguished, i.e., their foreground / background relationship in the image needs to be determined. The method is to take two points on the connecting edge and draw a line. The object whose area the line passes through is the foreground, and its z-value is assumed to be Z. The other object is the background, and its z-value is Z+1. If multiple object areas are adjacent, they need to be compared pairwise and their z-values ​​set accordingly. Finally, the foreground / background relationship is determined by the sorting of z-values; the smaller the z-value, the further forward the object is in the image, and the larger the z-value, the further back it is in the image.

[0077] S44. Remove these objects from the image to obtain a pure background image, with the z-axis of the background image set to reference 0.

[0078] In step S5 above, the specific contents include:

[0079] S51. Generate dynamic images of the extracted objects. (Introduction to generating a Stable-Diffusion model)

[0080] S52. For the objects identified in S3, the generation of the guiding text requires three parts: First, the noun of the identified object; second, the subject, action, and description obtained from the syntactic analysis in S2; and third, a preset template. If the target object matches the subject in the syntactic analysis of a sentence, the preset template is "Based on the subject's action and description, generate a series of images to complete the guiding text"; if the target object does not match the subject in the syntactic analysis, or if no verb indicating an action is found in the syntactic analysis, the preset template is "Based on the subject and description, find a suitable common action for the subject, such as: raising a hand, looking up, walking, jumping, etc., with a small range of motion. Complete the guiding text generated from the series of images based on this action."

[0081] In step S6 above, the specific contents include:

[0082] S61. Image generation is divided into three parts: static image completion, dynamic image generation of the foreground subject, and fusion of the dynamic foreground object and the static background image to generate the final dynamic image.

[0083] S62. Static image completion includes background completion or foreground objects being occluded by other objects, requiring completion. We input the background image or the occluded foreground image into a pre-trained image completion SD model to perform static image completion.

[0084] S63. Animated Graph Generation: Input the extracted object image, or the extracted and completed object image, and the corresponding prompts generated in step S5 into the pre-trained animated graph generation SD model to generate a series of multiple frames of images of the corresponding object and the prescribed action.

[0085] S64. Align the multi-frame images generated from multiple sets of objects along the timeline and stitch them into the completed static background image. During stitching, z-axis information needs to be used to maintain the overlapping relationship between the extracted objects.

[0086] S65. Generate a multi-frame image of continuous motion following the steps above. Play the multi-frame image continuously in a loop according to the set frame rate to create a page in a vivid animated picture book.

[0087] The image completion SD model and the animation generation SD model are two stable diffusion models. The former is specifically trained for completing picture book images and is an inpaiting model; the latter is a full image generation model. Both need to be used in conjunction with the prompts generated in step S5 to complete the image or to generate each frame of the animation.

[0088] In step S7 above, the specific contents include:

[0089] S71. For children's picture book content, pre-train the SwinBERT video summarization model.

[0090] S72. Input the dynamic video generated from a single page of a picture book into the model to obtain a descriptive summary of the video.

[0091] In step S8 above, the specific content includes,

[0092] S81. Speech synthesis and generation consists of three parts: one is to generate background audio based on video summaries, the second is to generate dialogue audio based on characters, and the third is to customize the voiceover audio.

[0093] S82. Based on the video summary obtained earlier, call the LLM model to query for the most suitable background sound type. Then, perform a matching search from the background sound library to obtain a suitable background sound.

[0094] S83. For children's picture books, pre-train VALL-E models for speech synthesis of various cartoon and animal characters. Based on the video summary content, if the objects or characters in the scene have dialogue, search the character speech database for matching character features (such as a certain animal or cartoon character) to find the best matching speech model. Synthesize the dialogue speech based on the text content extracted from the preceding picture book pages.

[0095] S84. If the text in the image is narration, a preset speech model can be used to generate the narration. Alternatively, the VALL-E model can be used to generate a customized speech model for the target from the input speech information (such as a mother's voice). This speech model can then be used to generate the narration.

[0096] S85. Sort the dialogue and narration in chronological order, then blend them with the background audio to generate the audio file for this page of the picture book. Combine the audio file with the previously created animated visuals to generate the final video file for this page of the picture book.

[0097] This invention provides an innovative, personalized, and highly interactive way to generate children's picture books, allowing children to access personalized picture book content and interact with images, text, and sounds, thus enhancing their reading and entertainment experience.

[0098] Based on the same inventive concept, this invention also provides a multimedia children's picture book generation system based on generative artificial intelligence. Since the principle of the problem solved by this system is similar to the aforementioned multimedia children's picture book generation method based on generative artificial intelligence, the implementation of this system can refer to the implementation of the aforementioned method, and the repeated parts will not be described again.

[0099] This invention provides a multimedia children's picture book generation system based on generative artificial intelligence, comprising:

[0100] The scanning and extraction module is used to acquire the scanned image of the target physical picture book, and to identify and extract the text information in the image.

[0101] The syntactic analysis module is used to analyze the syntax of the text information and extract the subject, action, and corresponding descriptive information. For example, text information in an image can be extracted through a text recognition model, and the sentence can be extracted using syntactic analysis to obtain the subject, action, and descriptive information.

[0102] The image object detection module is used to detect the main object in the image; it can identify and locate the main object in the image.

[0103] The image semantic segmentation module is used to separate the main object and the background in the image; it can extract the identified object from the background.

[0104] The introductory text generation module is used to generate introductory text for the Stable-Diffusion model based on the main object detected in the image, the subject, action and corresponding descriptive information extracted by analyzing the text information, and a preset template; the generated introductory text serves as the input for image generation of the Stable-Diffusion model.

[0105] The dynamic image generation module completes the background image and generates continuous dynamic images; specifically, it uses a pre-trained Stable-Diffusion model to generate dynamic images from the static images of the main objects separated from the image, and completes the background images of the separated main objects.

[0106] The video summary description module is used to input the dynamic image into the SwinBERT video summary module to obtain the corresponding video description summary;

[0107] The speech synthesis and generation module is used to synthesize corresponding narration, dialogue, or background audio based on the content of the dynamic images or pre-set content, thereby transforming static picture books into generative multimedia children's picture books. Background music is automatically matched based on video description summaries; dialogue or narration is generated by analyzing and extracting text from the video description summaries to produce relevant audio; timeline alignment and fusion merge the dynamic images and audio; thus transforming static picture books into generative multimedia children's picture books.

[0108] Multimedia children's picture book generation systems based on generative artificial intelligence have broad application prospects and economic benefits, and can play an important role in children's education, entertainment and publishing.

[0109] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0110] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multimedia children's picture book generation method based on generative artificial intelligence, characterized in that, Includes the following steps: S1. Obtain the scanned image of the target physical picture book, and identify and extract the text information in the image; S2. Analyze the syntax of the text information to extract the subject, action, and corresponding descriptive information; S3. Detect the main object in the image; S4. Separate the main object and the background in the image; S5. Based on the main object detected in the image, and the main body, action and corresponding description information extracted from the text information, combined with the preset template, generate the guiding text of the Stable-Diffusion model; S6. Using a pre-trained Stable-Diffusion model, generate a dynamic image from the static image of the main object separated from the image; and complete the background image of the separated main object. S7. Input the dynamic image into the SwinBERT video summarization module to obtain the corresponding video description summary; S8. Based on the content of the dynamic image or the pre-set content, synthesize the corresponding narration, dialogue or background sound, thereby transforming the static picture book into a generative multimedia children's picture book. Step S4 includes: The main object in the image is used as the foreground for semantic segmentation of the image; Obtain the mask pixels of the main object in the image, and extract the main object from the background; The z-axis information is added to the two-dimensional coordinates of the main object in the image to convert it into three-dimensional coordinates. If there is an overlap between the extracted objects, they are distinguished by the z-axis information. If the extracted regions of two objects are connected by an edge, two points are taken on the connecting edge and a line is drawn. The object whose region is crossed by the line is the foreground, and its z-axis value is Z. The other object is the background, and its z-axis value is Z+1. If multiple object regions are adjacent, the foreground and background are determined for each pair and the z-axis value is set accordingly. Finally, the foreground and background relationships are determined by the sorting of the z-axis values. The main object is removed from the image to obtain a background image; the z-axis of the background image is set to reference 0.

2. The multimedia children's picture book generation method based on generative artificial intelligence according to claim 1, characterized in that, Step S1 includes: Scan each page of the printed target physical picture book to create a digital image; or directly obtain each digital image of the electronic target physical picture book; The digital image is subjected to OCR text detection and recognition extraction, and the position of the text in the digital image is recorded.

3. The multimedia children's picture book generation method based on generative artificial intelligence according to claim 1, characterized in that, Step S2 includes: A pre-trained NLP syntactic analysis model is used to perform syntactic analysis on the text information, extracting the subject, action, and corresponding descriptive information from the descriptive sentences in the picture book.

4. The multimedia children's picture book generation method based on generative artificial intelligence according to claim 1, characterized in that, Step S3 includes: The object recognition model, pre-trained for children's picture books, is used to identify objects in the image. Identify the main object in the image and its position in the image.

5. The multimedia children's picture book generation method based on generative artificial intelligence according to claim 1, characterized in that, Step S5 further includes: If the main object matches the subject in the syntactic analysis of a sentence, the preset template is: generate a series of images based on the subject, action, and corresponding descriptive information to complete its introductory text; If the target object cannot match the subject in the syntactic analysis, or if no verb indicating an action is found in the syntactic analysis, the preset template is: based on the subject and the corresponding descriptive information, a preset action is set, and the guide text generated by the continuous images is completed based on the action.

6. The multimedia children's picture book generation method based on generative artificial intelligence according to claim 1, characterized in that, Step S6 includes: S61. Input the background image or the occluded foreground image separated from the image into the pre-trained Stable-Diffusion model to complete the static image. S62. Input the extracted main object image, or the extracted and completed main object image, and the guiding text into the pre-trained Stable-Diffusion model to generate a series of multi-frame images of the corresponding object and the corresponding action. S63. Align the multi-frame images generated from multiple groups of objects along the timeline and stitch them into the completed static background image; during stitching, use the z-axis information to maintain the overlapping relationship between the extracted objects. S64. Generate multi-frame images of continuous motion according to the above steps S61~S63; play the multi-frame images continuously in a loop according to the set frame rate to create a page in the dynamic picture book.

7. The multimedia children's picture book generation method based on generative artificial intelligence according to claim 1, characterized in that, Step S8 includes: Based on the obtained video description summary, the LLM model is invoked to obtain the corresponding background sound type; a matching search is performed from the background sound library to obtain the corresponding background sound; For children's picture books, VALL-E models for speech synthesis of various cartoon and animal characters are pre-trained; based on the video description summary, if the objects and characters in the scene have dialogue, the system searches the character speech database to find matching character features and finds the most matching speech model; based on the text content extracted by the syntactic analysis, the dialogue speech is synthesized. If the text in the image is narration, the narration is generated using a preset speech model; The dialogue and narration are sorted according to time sequence and then blended with the background sound to generate the audio file of the current page of the picture book; the audio file and the content of the dynamic images are combined to generate the final video file of the current page of the picture book.

8. A multimedia children's picture book generation system based on generative artificial intelligence, characterized in that, include: The scanning and extraction module is used to acquire the scanned image of the target physical picture book, and to identify and extract the text information in the image. The syntactic analysis module is used to analyze the syntax of the text information and extract the subject, action, and corresponding descriptive information. An image object detection module is used to detect the main object in the image; The image semantic segmentation module is used to separate the main object and the background in the image; The introductory text generation module is used to generate introductory text for the Stable-Diffusion model based on the main object detected in the image, the subject, action and corresponding descriptive information extracted by analyzing the text information, and a preset template. The dynamic image generation module is used to generate dynamic images from the static images of the main objects separated from the image using a pre-trained Stable-Diffusion model; and to complete the background images of the separated main objects. The video summary description module is used to input the dynamic image into the SwinBERT video summary module to obtain the corresponding video description summary; The speech synthesis and generation module is used to synthesize corresponding narration, dialogue or background sound based on the content of the dynamic image or the pre-set content, thereby transforming the static picture book into a generative multimedia children's picture book. Specifically, the image semantic segmentation module is used for: The main object in the image is used as the foreground for semantic segmentation of the image; Obtain the mask pixels of the main object in the image, and extract the main object from the background; The z-axis information is added to the two-dimensional coordinates of the main object in the image to convert it into three-dimensional coordinates. If there is an overlap between the extracted objects, they are distinguished by the z-axis information. If the extracted regions of two objects are connected by an edge, two points are taken on the connecting edge and a line is drawn. The object whose region is crossed by the line is the foreground, and its z-axis value is Z. The other object is the background, and its z-axis value is Z+1. If multiple object regions are adjacent, the foreground and background are determined for each pair and the z-axis value is set accordingly. Finally, the foreground and background relationships are determined by the sorting of the z-axis values. The main object is removed from the image to obtain a background image; the z-axis of the background image is set to reference 0.

Images