JP2009130899A - Image playback apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image playback apparatus capable of accessing a telephone or Web by extracting, as text data, a telephone number, URL, retrieval word, and the like from a moving image of one-segment broadcasting or the like. <P>SOLUTION: An image playback apparatus includes: a character region extracting means 3 for extracting, as a character region image, a region containing characters from input image information; a character region image accumulating means for storing the character region image extracted by the character region extracting means 3; a character region processing means 5 for processing characters contained in the character region image; a character recognizing means 6 for recognizing characters from the character region image processed by the character region processing means 5 to produce character text data; a character information accumulating means 7 for storing the character text data produced by the character recognizing means 6; a key input means 9 for performing input operation from the outside; and a control means 8 for controlling respective operational sections based on input instructions of the key input means 9. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image reproducing apparatus that extracts text data from a moving image and can directly access the outside from the extracted text data.

  In recent years, many mobile phones have been proposed that can watch TV broadcasts such as one-segment broadcasting and those that have a full browsing function. In one-segment broadcasting, in particular, a commercial message (hereinafter referred to as CM) portion often includes information such as a telephone number such as a free dial, a URL (Uniform Resource Locator) on the Internet, and a search word. Viewers and users who are interested in the CM need to store this information and access the telephone and the Web (World Wide Web) in another step.

  As a means of recording information in a conventional mobile phone, for example, a business card or timetable is photographed with a digital camera built in an image playback device such as a mobile phone, and the information is stored and utilized as image format data. There is something that is. In addition, for example, data in an image format captured by a digital camera built in a mobile phone is stored as a text format database in combination with optical character recognition (OCR), so that the stored information can be stored. There has been proposed an apparatus and a method that can be easily updated when changed (see, for example, Patent Document 1).

JP 2007-164785 A

  Patent Document 1 describes recording information using a digital camera built in a mobile phone, and does not describe recording information displayed during one-segment broadcasting.

  It is often difficult for a user to accurately record a telephone number, URL, and search word. Also, when watching 1Seg broadcasting with a mobile phone in a moving body such as a train, it is difficult to read due to the shaking of the moving body and the low resolution of the information displayed on the screen. It often ends up.

  The present invention has been made to solve these problems, and by reproducing a telephone number, a URL, a search word, and the like as text data from a moving image such as one-segment broadcasting, image reproduction that can be accessed on the telephone or the Web. An object is to provide an apparatus.

  In order to solve the above-described problems, an image reproduction apparatus according to the present invention includes a character area extraction unit that extracts a region including a character from the input image information as a character region image, and a character extracted by the character region extraction unit. Character area image storage means for storing area images, character area processing means for processing characters included in the character area image, and character text data by recognizing characters from the character area image processed by the character area processing means A character recognizing means to be generated, a character information accumulating means for storing character text data generated by the character recognizing means, a display means for displaying an operation menu relating to image information, character text data or character area image, and externally It is characterized by comprising key input means for performing an input operation, and control means for controlling each operation unit based on an input instruction of the key input means.

  According to the present invention, the character area extraction unit extracts a region including characters from the input image information as a character region image, the character region image storage unit stores the character region image extracted by the character region extraction unit, The character area processing means processes characters included in the character area image, the character recognition means recognizes characters from the character area image processed by the character area processing means, generates character text data, and the character information storage means Since character text data generated by the character recognition means is stored, a telephone number, a URL, a search keyword, and the like are extracted as text data from a moving image such as one-segment broadcasting, thereby enabling access to the telephone or the Web.

  Embodiments of the present invention will be described below with reference to the drawings.

<Embodiment 1>
FIG. 1 is a block diagram showing a configuration of an image reproduction apparatus 101 according to Embodiment 1 of the present invention. As shown in FIG. 1, an image reproducing device 101 includes a frame memory 1 that records input image data (image information) for each frame, and a CM period detector 2 (CM period detector) that detects a CM period. A character region extraction unit 3 (character region extraction means) that extracts a region including characters from the input image data as a character region image, and a character region that stores the character region image extracted by the character region extraction unit 3 Characters from the image storage unit 4 (character region storage unit), the character region processing unit 5 (character region processing unit) that processes characters included in the character region image, and the character region image processed by the character region processing unit 5 A character recognition unit 6 (character recognition means) that generates character text data (text data) and a character information storage unit 7 (sentence) that stores character text data generated by the character recognition unit 6 An information storage unit), a display unit 10 (display unit) for displaying image data and an operation menu, a key input unit 9 (key input unit) for performing an input operation from the outside, and an input instruction from the key input unit 9 It comprises a control unit 8 (control means) that controls each operation unit.

  The feature of Embodiment 1 of the present invention is that, while image data is being input, the control unit 8 selects and selects text data stored in the character information storage unit 7 based on an input instruction from the key input unit 9. Phone connection using a telephone number included in the text data, and Internet connection using an address or keyword.

  The frame memory 1 is composed of a memory such as SDRAM, and records image data input after decoding video content encoded by MPEG4 (Moving Picture Experts Group 4) or the like in units of one frame. The image data read from the frame memory 1 is sequentially input to the CM period extraction unit 2.

  The CM period detection unit 2 detects a CM period that is between broadcast programs by detecting a silent period that is present when switching to stereo broadcasting, which is a characteristic of CM, or when a broadcast program is changed to CM. In addition, detection of silent periods inserted for each CM, count of CM periods for 15 seconds or 30 seconds, results of detecting video change (scene change) from changes in histograms between 2-frame image data, etc. When used together, a more accurate CM period can be detected. CM video, which is image data detected by the CM period detection unit 2, is read from the frame memory 1 and input to the character region extraction unit 3.

  In the embodiment of the present invention, since the telephone number, URL, and search word displayed on a moving image such as a broadcast program are often displayed mainly during the CM period, the CM period detection unit 2 detects them. Only the CM video that has been processed is output to the character area extraction unit 3, but when the search word displayed in a program other than the CM is detected and used, the processing by the CM period detection unit 2 is omitted. Also good.

  Next, the character area extraction unit 3 will be described in detail.

  The character region extraction unit 3 extracts a region including characters from the input image data as a character region image. Characters in the character area image are characterized by little change or movement. For this reason, the frame of the input image data is compared with the frame of the image data input immediately before, and there is a high possibility that an area with little or no change is a character area image.

As a method of extracting the character area image, the character area image is limited using an index of character character. What is the character quality indicator?
-The frequency component is high, that is, the number of edges detected in a region obtained by dividing the input image data is large.
-The same pixel value continues for the thickness of the character line. That is, the pixel values in the outline of the character are substantially uniform.
-When rectangle processing is performed on the outline of a character, the rectangles are aligned at regular intervals.
and so on.

  FIG. 2 is a block diagram showing the configuration of the character area extraction unit 3 according to Embodiment 1 of the present invention. As shown in FIG. 2, the character region extraction unit 3 includes an edge extraction unit 301, a binarization unit 302, a binary information holding unit 303, an edge stationary detection unit 304, an edge motion detection unit 305, a character likelihood determination unit 306, It consists of a static area holding unit 307, a movement area holding unit 308, a character area input detection unit 309, a character area disappearance detection unit 310, a character area detection unit 311, and a labeling / rectangular shaping unit 312.

  The edge extraction unit 301 extracts edge information included in the character area image by using a spatial filter such as a Sobel filter or a Prewitt filter.

  The edge information extracted by the edge extraction unit 301 is expressed by the binarization unit 302 as binary values of white and black. At this time, for example, the edge may be white (1) and the other may be black (0). As a result of the binarization, the information amount of the character area image is reduced, and subsequent processing is simplified.

  In the binarization method, for example, there are many emphasized characters used in CM or the like as target characters in the present invention, and these characters are binarized with a predetermined pixel value because they have high brightness and high contrast. It is also possible to use a fixed threshold value for performing. In addition, when the character does not have a feature of high brightness or high contrast compared to the background, an automatic binary that obtains an average of pixel values in the vicinity of the pixel of interest and binarizes it from the magnitude relationship with the average value. The optimal binarization method varies depending on the characteristics of the input image. For example, the binarization method can be selectively used. Ideal to apply.

  The binary information holding unit 303 holds binary information of one frame before the input image data.

  The edge stationary detection unit 304 compares the binary information of the current frame in the binarizing unit 302 with the binary information of the previous frame in the binary information holding unit 303, and detects that the edge is stationary. . For example, for each pixel unit in the frame, a portion where both the binary information of the binarizing unit 302 is white (1) and the binary information of the binary information holding unit 303 is white (1) is detected as an edge. To do. As a detection result, a flag indicating stationary edge information is expressed as a binary value, 1 otherwise, and 0.

  The edge motion detection unit 305 compares the binary information of the current frame in the binarization unit 302 with the binary information of the previous frame in the binary information holding unit 303 and detects that the edge has moved. For example, a pixel in which the binary information in the binary information holding unit 303 is white (1) is detected as a pixel in which the binary information in the binarizing unit 302 is black (0). Alternatively, a pixel in which the binary information in the binary information holding unit 303 is black (0) is detected as a pixel in which the binary information in the binarization unit 302 is white (1). Alternatively, the binary information of the binary information holding unit 303 and the binarizing unit 302 is compared, and a portion where white (1) is changed to black (0) and black (0) is changed to white (1) is detected. . As a detection result, a flag indicating edge information that has moved is represented as a binary value, and otherwise, it is represented as a binary value.

  The character-likeness determination unit 306 divides the region into, for example, 4 pixels × 4 pixels, and uses the fact that there are a large number of edges and a high frequency in the region where characters exist, The character-likeness is determined by the percentage of the moving edge (greater than a predetermined threshold). A flag indicating a stationary character area image and a moving character area image is added to each divided area.

  In the embodiment of the present invention, the number of edges in the divided area is large as a character-like index, but the above-described method may be used as another character-like index. In the embodiment of the present invention, the stationary and moving edges are detected by comparing the binary information of the binarizing unit 302 and the binary information holding unit 303, and the character likelihood determining unit 306 is detected based on the detection result. The character likelihood is determined by the character likelihood determination unit 306 based on the information binarized by the binarization unit 302 and the binary information holding unit 303, and the character is a determination result. Edge stationary and motion may be detected in the region.

  The still area holding unit 307 and the moving area holding unit 308 hold flags indicating the stationary character area and the moving character area added by the character likelihood determination unit 306, respectively.

  The character area input detection unit 309 uses the information that the current edge is stationary and the information that the edge has moved in the past that is held in the movement area holding unit 308 to the character data in the input image data. Detect that information has been inserted. For example, in a character area image immediately after character information is inserted, such as a telop, there is movement at the edge, and then the edge is stationary.

  The character area disappearance detection unit 310 obtains character information from the input image data from the information that the current edge has moved and the information that the edge has been stationary in the past held in the static region holding unit 307. Detect that it has disappeared. For example, in the character area image immediately before the character information such as telop disappears, the edge is stationary and then the edge moves.

  The character area detection unit 311 detects, as a character area image, an area where an edge is stationary between a plurality of predetermined frames and is determined to be a character. The character area image detected by the character area detection unit 311 may be erroneously detected in a background image having similar characteristics. In order to improve this erroneous detection, the output from the character area input detection unit 309 is held and accumulated and used as a mask signal. Using this mask signal, the area other than the character area image for which the character area is determined to be inserted by the character area input detection unit 309 is excluded from character detection. On the other hand, the character area image is updated by resetting the mask signal by the output from the character area disappearance detection unit 310.

  The output of the character area image address or the image information of the character area image from the character area detection unit 311 is performed at the character area information update timing that is the output of the character area disappearance detection unit 310. This is because the same or similar information is not output to the subsequent character area image storage unit 4.

  In addition, when there is a margin in the storage capacity of the character area image storage unit 4, the output is not limited to the above timing. In addition, the character area detection unit 311 detects an area that is determined to be a character by detecting stillness of a predetermined edge between a plurality of frames after the character area is inserted based on the output of the character area input detection unit 309. It may be detected as a region image. At this time, the character region information update timing may be a predetermined frame after the output of the character region input detection unit 309.

  In the labeling / rectangular shaping unit 312, expansion and contraction processing is performed to fill in the missing tooth portion of the missing tooth outputted from the character region detecting unit 311 and to remove the isolated region. Furthermore, a process such as adding identification information by applying a labeling process to the character area image, and narrowing down the target character string by adjusting the area, the aspect ratio of the rectangle, etc. for each added identification information To do.

  With the above processing, the character area image is designated only from the start point address and the end point address. Thereafter, the character area image is output to the character area image storage unit 4 via the frame memory 1.

  The storage of the character region image from the character region extraction unit 3 to the character region image storage unit 4 may be performed when the disappearance of the character region image is detected by the character region disappearance detection unit 310. Alternatively, it may be performed when a character region image insertion is detected by the character region input detection unit 310, or after a predetermined image frame after the insertion of the character region image is detected.

  Since character recognition in the character recognition unit 6 is performed by pattern matching with a general font dictionary or the like, high-resolution characters are required. In addition, it is required that the character and the background can be easily distinguished, such as black for the character and white for the background. In many cases, the character cannot be recognized by a special font such as a border character. Therefore, the character region processing unit 5 performs image processing on the character region image stored in the character region image storage unit 4 to improve the quality of the character image.

  Next, the character area processing unit 5 will be described in detail.

  FIG. 3 is a block diagram showing the configuration of the character area processing unit 3 according to Embodiment 1 of the present invention. As shown in FIG. 3, the character area processing unit 5 includes a character size determination unit 501, an enlargement processing unit 502, a binarization unit 503, an equalization processing unit 504, an edge enhancement unit 505, a brightness / contrast adjustment unit 506, a blur. It consists of a removal unit 507.

  The character size determination unit 501 determines the actual character size by detecting that there is no edge between characters and between character strings (lines). For example, when the character area image is composed of one line of character string and when it is composed of three lines, the enlargement ratio when the enlargement processing unit 502 performs enlargement processing is affected, so that the character size is grasped. is required. Specifically, the divided area is made smaller than when the character area image is determined, thereby detecting a missing portion in the rectangular area.

  The enlargement processing unit 502 requires a high-performance enlargement process to cope with the fact that the character recognition unit 6 requires high resolution. As a data interpolation method, a countermeasure against a jaggedness (jaggy) of a contour that is particularly likely to occur in an oblique line portion is included in an algorithm such as a cubic interpolation (Bancubic, Lancoz 3) method.

  The processing in FIG. 3I is performed when the background and characters of the character region image extracted by the character region extraction unit 3 are each in a single color or similar, and the background and characters can be easily separated by binarization. To be done. Whether or not binarization is easy for a character area image can be determined by examining a histogram relating to each pixel value of the character area image.

  The processing in FIG. 3II is performed when it is not easy to separate the background and the characters. The equalization processing unit 504 averages the histogram relating to the brightness of the character area image to create a once blurred image. Thereafter, the edge enhancement unit 505 emphasizes the edge portion. For example, when the background is white and the characters are black, the brightness / contrast adjustment unit 506 can separate the image from the background image by converting the image to a high-contrast image and then adjusting the brightness to a bright image.

  In the processing in FIG. 3III, binarization is performed by the binarization unit 503 after the processing in FIG. 3II, and the background of the character area image and the characters are more clearly separated. The processing in the binarization unit 503 in FIG. 3I is also the same. However, when binarization is performed, there are many cases in which image deterioration occurs, such as blurring of characters. Make the line width uniform.

  As described above, the character area image subjected to image processing by the character area processing unit 5 is output to the character recognition unit 6.

The character recognition unit 6 performs character recognition by performing pattern matching on a font dictionary or the like. Since character strings such as phone numbers, URLs, and search words displayed during one-segment broadcasting commercials lack resolution, as a method for recognizing such character strings,
• Optimize the pattern matching method.
・ Limit the target character string to numbers or English letters (alphabet).
-Lower the character level (kanji level) included in the dictionary.
Optimize by such as. Character information that has been recognized by the character recognition unit 6 and extracted as text data is stored in the character information storage unit 7. In addition, when the character recognition is completed by the character recognition unit 6, it may be configured to immediately notify the user by displaying “character recognition completed!” Or the like.

  The characters extracted from the one-segment broadcast video may include characters that are not recognized 100% correctly by the character recognition unit 6 and are erroneously recognized. As described above, since the character area image stored in the character area image storage unit 4 is a result of processing in the character area extraction unit 3, for example, the character area image is converted into the character area input detection unit 309 of the character area extraction unit 3. When the character region image storage unit 4 stores the same information in accordance with the character region information update timing output from each of the character region disappearance detection unit 310, two pieces of the same information are stored. However, the character region information update timing from the character region input detection unit 309 and the character region information update timing from the character region disappearance detection unit 310 are different in the frame of the image data and the background image changes. Either one of these may not be detected accurately. Therefore, as long as processing time and memory capacity allow, performing character recognition from a plurality of frame data is a desirable method from the viewpoint of improving character recognition accuracy.

In the embodiment of the present invention, frame images are read from the frame memory 1 at regular intervals in parallel with video playback such as one-segment broadcasting on the display unit 10, and phone numbers, URLs, and search words are extracted and characters are recognized. Assuming that In the CM, a character area image is always extracted and character recognition is performed, and character information is stored in the character information storage unit 7 as text data.
Phone number: numbers only, starting with 0120-
・ URL: Most of alphabet letters, www and. com.
Search word: By analyzing character information such as a character string other than the above, it can be classified into, for example, a telephone number, a URL, and a search word.

  Next, an operation when a user who views a broadcast program selects a keyword to be accessed will be described.

  FIG. 4 is a schematic diagram illustrating a user operation according to the first embodiment of the present invention. The user operates the operation menu displayed on the display unit 10 via the key input unit 9. Here, the operation menu displays information related to character text data or character region images.

  In FIG. 4A, the user presses the “keyword link” button A when accessing the telephone or the Web with interest in the CM being watched or watched. By pressing the left / right button B, one that is classified into “phone number”, “URL”, and “search word” is selected according to the purpose of the user. For example, from the state of “telephone number” in FIG. 4A, the “search word” as shown in FIG. 4B is selected and displayed by pressing the left / right button B.

  At this time, when the control unit 8 detects that the “keyword link” button A is pressed, the telephone number is read from the text data stored in the character information storage unit 7 and the display on the display unit 10 is changed to “phone number”. And When it is detected that the user has selected the “search word” by pressing the left / right button B, the search word is read from the text data stored in the character information storage unit 9 and the display of the display unit 10 is changed to “search word”. ".

  In FIG. 4B, the user selects a desired character string from the “search word” list by pressing the up / down button C.

  In FIG. 4C, a message prompting confirmation such as “Is the selected keyword correct?” Is displayed on the display unit 10 together with a list of “search words”. If there is no error in the selected character string, “Yes” is selected and determined. At this time, if “Yes” is selected and determined from the key input unit 9, the control unit 8 accesses the Web via a browser search engine. In the case of “telephone number”, the telephone connection is made, and in the case of URL (address), the Web is directly accessed.

  If there is an error in the phone number or the search word displayed on the display unit 10 due to erroneous recognition of characters, the user can select the phone number or the search word by selecting “No” with the key input unit 9 and determining it. It can be corrected.

  Thereafter, the user corrects himself / herself at the correction input portion D in FIG. 4D, and when the correction is completed, the user presses the “link” key for accessing the telephone or the Web, and presses the decision key E, thereby controlling the control unit 8. Accesses the Web.

  When the user modifies the search word by himself / herself, by displaying the original image of the character area as shown in FIG. 4D, the user confirms the characters to be input while viewing the original image when correcting can do. Further, as shown in FIG. 5E, when correcting a telephone number, it is impossible to determine which telephone number is based on a simple enumeration of numbers, so the frame image on which the selected telephone number is displayed is reduced. indicate.

  As shown in FIG. 5F, the original image of the character region and the reduced image of the frame may be displayed simultaneously when the user selects a character string to be accessed from the list. Although not shown, for the same purpose, the user may be able to listen to the voice of the CM on which the selected telephone number is displayed.

From the above, the phone number, URL, and search word can be stored as text data by automatically extracting character areas and recognizing characters from broadcast programs such as commercials. By selecting, it becomes possible to access the telephone and the Web. Even if there is an error in the text data stored by character recognition, correction is possible and easy by associating it with the original image on which the text data is displayed.

<Embodiment 2>
FIG. 6 is a block diagram showing the configuration of the image playback device 101a according to the second embodiment of the present invention. In the second embodiment, the process of the control unit 8a is different from that of the first embodiment, and the other configurations and processes are the same, and thus the description thereof is omitted here. In the image reproduction device 101a of the second embodiment, the character area image stored in the character area image storage unit 4 can be displayed on the display unit 10.

  The feature of Embodiment 2 of the present invention is that the control unit 8a selects a desired character region image from the character region image storage unit 4 based on an input instruction of the key input unit 9, and the character region processing unit 5 selects the character region image. And text data obtained by processing by the character recognition unit 6, and telephone connection using a telephone number included in the text data, and Internet connection using an address or a keyword.

  FIG. 7 is a schematic diagram illustrating a user operation according to the second embodiment of the present invention. The user operates the operation menu displayed on the display unit 10 via the key input unit 9. Here, the operation menu displays information related to character text data or character region images.

  In FIG. 7A, the user presses the “keyword link” button A when accessing the telephone or the Web with interest in the CM being watched or watched.

  At this time, when the control unit 8a detects that the “keyword link” button A is pressed, the character region image stored in the character region image storage unit 4 is read, and the read character region image is “keyword character string”. As a list on the display unit 10.

  The user selects a character string to be accessed from the list of “keyword character strings” displayed on the display unit 10 by using the up / down button B, and presses the decision key E in FIG.

  When the user selects a character string to be accessed via the key input unit 9, the control unit 8 a reads out the target character region image from the character region image storage unit 4 and outputs it to the character region processing unit 5. Then, after image processing is performed by the character region processing unit 5, the character recognition unit 6 recognizes the character and stores it once in the character information storage unit 7. When the information stored in the character information storage unit 7 is transmitted to the control unit 8a, the control unit 8a displays a message prompting confirmation such as “Is the selected keyword correct?” On the display unit 10 (not shown). ) Further, the control unit 8a analyzes the text data stored in the character information storage unit 7 and determines whether it is a telephone number, a URL, or a search word.

  The subsequent processing is the same as in the first embodiment. That is, if “Yes” is selected, the telephone or the web is accessed, and if “No” is selected, the user accesses the telephone or the web after making corrections by himself / herself.

  From the above, after the user selects a character area image stored in the character area image storage unit 4 as a keyword, only the selected keyword is processed and recognized, thereby reducing the time required for the process. Is possible.

It is a block diagram which shows the structure of the image reproduction apparatus by Embodiment 1 of this invention. It is a block diagram which shows the structure of the character area extraction part by Embodiment 1 of this invention. It is a block diagram which shows the structure of the character area process part by Embodiment 1 of this invention. It is a schematic diagram which shows a user's operation by Embodiment 1 of this invention. It is a schematic diagram which shows a user's operation by Embodiment 1 of this invention. It is a block diagram which shows the structure of the image reproduction apparatus by Embodiment 2 of this invention. It is a schematic diagram which shows a user's operation by Embodiment 2 of this invention.

Explanation of symbols

  101 image playback device, 101a image playback device, 1 frame memory, 2 CM period detection unit, 3 character region extraction unit, 4 character region image storage unit, 5 character region processing unit, 6 character recognition unit, 7 character information storage unit, 8 control unit, 8a control unit, 9 key input unit, 10 display unit, 301 edge extraction unit, 302 binarization unit, 303 binary information holding unit, 304 edge stationary detection unit, 305 edge motion detection unit, 306 character character Determination unit, 307 Still area holding unit, 308 Motion area holding unit, 309 Character area input detection unit, 310 Character area disappearance detection unit, 311 Character area detection unit, 312 Labeling / rectangular shaping unit, 501 Character size determination unit, 502 Enlarge Processing unit, 503 binarization unit, 504 equalization processing unit, 505 edge enhancement unit, 506 brightness / contrast adjustment Department, 507 blur removal unit.

Claims (7)

A character region extraction means for extracting a region including characters from the input image information as a character region image;
A character area image storage means for storing the character area image extracted by the character area extraction means;
Character region processing means for processing characters included in the character region image;
Character recognition means for recognizing characters from the character area image processed by the character area processing means and generating character text data;
Character information storage means for storing the character text data generated by the character recognition means;
Display means for displaying an operation menu relating to the image information, the character text data or the character region image;
A key input means for performing an input operation from the outside;
Control means for controlling each operation unit based on an input instruction of the key input means;
An image reproducing apparatus comprising: CM period detecting means for detecting a CM (commercial message) period is further provided.
2. The image reproducing apparatus according to claim 1, wherein extraction of the character area image by the character area extracting unit is limited to the CM period.   While the image information is input, the control unit selects the character text data stored in the character information storage unit based on an input instruction from the key input unit, and is included in the selected character text data. 3. The image reproducing apparatus according to claim 1, wherein a telephone connection by a telephone number and an Internet connection by an address or a keyword are performed.   The control means selects a desired character area image from the character area image storage means based on an input instruction of the key input means, and processes the character area image by the character area processing means and the character recognition means. The image reproducing apparatus according to claim 1 or 2, wherein the character text data is obtained, and a telephone connection using a telephone number included in the character text data and an Internet connection using an address or a keyword are performed.   2. The character area image stored in the character area image storage unit and the character text data stored in the character information storage unit are associated with each other and displayed on the display unit. The image reproducing device according to claim 4. The character area extraction means includes character area disappearance detection means for detecting that the character area image has disappeared from the input image information.
6. The storage of the character area image in the character area image storage means is performed when the disappearance of the character area image is detected by the character area disappearance detection means. The image reproducing device according to any one of the above. The character area extraction means includes character area input detection means for detecting that the character area image has been inserted into the input image information.
The character area image is stored in the character area image storage means when the insertion of the character area image is detected by the character area input detection means or after the insertion of the character area image is detected. 6. The image reproducing apparatus according to claim 1, which is performed later.

Images