Barcode image recognition method and device

Abstract

A bar code image recognition device comprises a camera module, an input element, a processing module, a storage module, an output module and a display. The input element is coupled to the camera module, the processing module is coupled to the camera module, and the processing module comprises a first artificial neural network, a conversion module, a second artificial neural network and a comparison module. Wherein the first artificial neural network identifies a first target image of each target object in a preset image shot by the camera module, and the second artificial neural network identifies a second target image in an intercepted image shot by the camera module. The storage module is coupled to the processing module. The output module is coupled to the comparison module and the storage module. The display is coupled to the output module.

Description

Technical Field

[0001] The present invention relates to a barcode recognition technology, and in particular to a method and device for recognizing a barcode image. Background Art

[0002] Barcodes can carry multi-character information, such as product name, specifications, origin, price, batch number and other related information, and are often used to mark goods, track logistics, manage inventory and control production processes. In these logistics, management and control, accurate and controllable recognition of barcodes or text can improve the operating efficiency of relevant staff and reduce manual operations and errors. Summary of the invention

[0003] In view of this, an object of the present invention is to provide a barcode image recognition method and device thereof, which is suitable for accurately and controllably outputting a string of indicated barcodes or characters when there are multiple barcodes or characters at the same time.

[0004] In one embodiment, a barcode image recognition method includes: starting a camera module; after the camera module is started, using the camera module to continuously pre-shoot at least one target object to obtain a plurality of preset images with continuous shooting time; using a first artificial neural network to identify a first target image of each target object in each preset image; extracting each first target image from each preset image; converting each first target image into a string; storing each first target image and the string corresponding to each first target image; triggering the camera module to shoot a single shot of a predetermined mark according to a confirmation signal to obtain a captured image; using a second artificial neural network to identify a mark image of a predetermined mark in the captured image and a second target image of a selected target object overlapping with the predetermined mark in at least one target object; comparing the second target image with each first target image to obtain a selected image, the selected image being the first target image with the highest similarity to the second target image; and reading and outputting the string corresponding to the selected image from the stored string.

[0005] The present invention further provides a barcode image recognition device, comprising a camera module, an input element, a processing module, a storage module, an output module and a display. The camera module is used to continuously pre-shoot at least one target object to obtain a plurality of preset images with continuous shooting time. The input element is coupled to the camera module to generate a confirmation signal to trigger the camera module to shoot a single shot of a predetermined mark to obtain a captured image in which the predetermined mark overlaps with at least one target object. The processing module is coupled to the camera module, comprising a first artificial neural network, a conversion module, a second artificial neural network and a comparison module. The first artificial neural network identifies the first target image of each target object in each preset image. The conversion module converts each first target image into a string. The second artificial neural network identifies the mark image of the predetermined mark in the captured image and the second target image of the selected target object overlapping with the predetermined mark in at least one target object; the comparison module is coupled to the first artificial neural network and the second artificial neural network, and compares the second target image with each first target image to obtain a selected image, and the selected image is the first target image with the highest similarity to the second target image. The storage module is coupled to the processing module, and the storage module stores each first target image and the character string corresponding to each first target image. The output module is coupled to the comparison module and the storage module, and the output module reads the character string corresponding to the selected image from the storage module. The display is coupled to the output module, and the display displays the character string read by the output module.

[0006] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments, but is not intended to limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1A is a functional block diagram of a barcode image recognition device according to some embodiments.

[0008] Figure 1B for Figure 1A A schematic diagram of an application scenario of the barcode image recognition device of an embodiment.

[0009] Figure 2 The present invention is a flowchart of a barcode image recognition method according to some embodiments.

[0010] Figure 3 for Figure 2 FIG. 1 is a schematic diagram of a preset image in step S110 of FIG.

[0011] Figure 4 for Figure 1A A functional block diagram of an exemplary conversion module in an embodiment.

[0012] Figure 5 for Figure 2 Flow chart of the barcode image recognition method of some embodiments after step S120.

[0013] Fig. 6AFIG. 4 is a schematic diagram of an application scenario of a barcode image recognition device according to another embodiment.

[0014] Figure 6B for Fig. 6A Functional block diagram of the barcode image recognition device in .

[0015] Figure 7 for Figure 6B A schematic diagram of an exemplary barcode image recognition device.

[0016] Figure 8 for Figure 6B Schematic diagram of another exemplary embodiment of a barcode image recognition device.

[0017] Fig. 9 for Figure 2 Flow chart of the barcode image recognition method of some embodiments before step S150.

[0018] Fig.10 for Figure 6B A schematic diagram of an exemplary barcode image recognition device.

[0019] Among them, the reference numerals

[0020] 10: Target

[0021] 122,124: Barcode

[0022] 146,148: string

[0023] 16: Selected Target

[0024] 20: Camera module

[0025] 22: Preset Image

[0026] 220: First target image

[0027] 224: Barcode Image

[0028] 226: String Image

[0029] 24: Capture images

[0030] 240,242: Labeled images

[0031] 26: Second target image

[0032] 30: Input components

[0033] 32: Button

[0034] 40: Processing module

[0035] 42: The First Artificial Neural Network

[0036] 44:Conversion module

[0037] 440:Decoder

[0038] 442: Identifier

[0039] 450,452: string

[0040] 46: Second Artificial Neural Network

[0041] 48: Comparison module

[0042] 480: Select image

[0043] 49:Mark confirmation module

[0044] 50: Indicator

[0045] 52: Established Mark

[0046] 60: Storage module

[0047] 70: Output module

[0048] 80: Display

[0049] 82: Display column

[0050] 90: Portable electronic devices

[0051] 92: Processor

[0052] S100~S190: Steps DETAILED DESCRIPTION

[0053] The structural principle and working principle of the present invention are described in detail below in conjunction with the accompanying drawings:

[0054] refer to Figure 1A The barcode image recognition device comprises a camera module 20, an input element 30, a processing module 40, a storage module 60, an output module 70 and a display 80. The input element 30 and the processing module 40 are respectively coupled to the camera module 20, the storage module 60 is coupled to the processing module 40, the output module 70 is coupled to the processing module 40 and the storage module 60, and the display 80 is coupled to the output module 70.

[0055] The processing module 40 includes a first artificial neural network 42, a conversion module 44, a second artificial neural network 46 and a comparison module 48. The first artificial neural network 42 receives the image captured by the camera module 20, the conversion module 44 is coupled to the first artificial neural network 42 and processes the image captured by the camera module 20, the second artificial neural network 46 also receives the image captured by the camera module 20, and the comparison module 48 is coupled to the first artificial neural network 42 and the second artificial neural network 46 and also processes the aforementioned images.

[0056] refer to Figure 1B , Figure 2 When using the barcode image recognition device, the camera module 20 is activated (step S100), and the camera module 20 continuously pre-shoots the target 10 to obtain a plurality of preset images 22 with continuous shooting time (step S110). In some embodiments, the target 10 is a barcode 122, 124 or a string 146, 148.

[0057] For reference Figure 1B , Figure 2 and Figure 3 The preset image 22 can be similar to the real-time preview of the captured image by the user through the screen when shooting with an electronic device. The preset image 22 has a first target image 220 of each target object 10. For example, if the target object 10 has a barcode 124 and a string 146, the preset image 22 will include a barcode image 224 of the barcode 122 and a string image 226 of the string 146. The barcode image 224 and the string image 226 are obtained by the first artificial neural network 42 in the processing module 40 recognizing the preset image 22 (step S120).

[0058] After the processing module 40 captures these identified first target images 220, the conversion module 44 converts each first target image 220 into a string 450, 452 (as can be seen in Figure 7 or Figure 8 )(Step S130).

[0059] refer to Figure 4 In some embodiments, the conversion module 44 may be a decoder 440. When converting the first target image 220 into a string 450, the decoder 440 decodes the barcodes 122, 124 to obtain information carried by the barcodes 122, 124, which may be text or a URL.

[0060] In some other embodiments, the conversion module 44 may be a recognizer 442 , and when converting the first target image 220 into the word string 452 , the recognizer 442 performs text recognition on the word strings 146 , 148 , and recognizes the word string image 226 as the word string 452 .

[0061] refer to Figure 4 and Figure 5 In some other embodiments, when the conversion module 44 includes both the decoder 440 and the identifier 442, the first artificial neural network 42 can identify the type of the first target image 220 (the type is the barcode image 224 or the string image 226 mentioned above).

[0062] Specifically, after the first artificial neural network 42 recognizes the first target image 220 in the preset image 22 (step S120), the first artificial neural network 42 further recognizes that each first target image 220 is a barcode image 224 or a string image 226 (step S122). If the first artificial neural network 42 recognizes that the first target image 220 is a barcode image 224 (step S125), the decoder 440 in the conversion module 44 decodes the barcode image 224 into a string 450 (step S132). If the first artificial neural network 42 recognizes that the first target image 220 is a string image 226 (step S127), the identifier 442 in the conversion module 44 performs text recognition on the string image 226 to obtain a string 452 (step S134).

[0063] Re-reference Figure 2 The storage module 60 stores the first target images 220 and the character strings 450, 452 corresponding to the first target images 220 obtained by the conversion module 44 (step S140). In some embodiments, the first target images 220 and the corresponding character strings 450, 452 in the plurality of preset images 22 are stored to form a database of barcodes 122, 124 and character strings 146, 148, so that the character strings 450, 452 corresponding to the barcodes 122, 124 and character strings 146, 148 can be retrieved at any time when needed.

[0064] Re-reference Figure 1B When the user wishes to obtain information about a specific target 10, the user can directly point to the specific target 10 through the pointing device 50 or a finger, thereby generating a predetermined mark 52 overlapping the target 10 on the target 10, and the indicated target 10 is used as the selected target 16. In some embodiments, the pointing device 50 is an infrared pen or a laser pen, and the predetermined mark 52 generated is a cursor; in some embodiments, the pointing device 50 can be implemented by a finger, and the predetermined mark 52 is the part of the target 10 pointed by the finger.

[0065] refer to Figure 1B , Figure 2 , Figure 7 and Figure 8After the user points to the target object 10 with the pointing device 50 or a finger, the camera module 20 takes a picture of the selected target object 16 according to a confirmation signal (step S150) to obtain a captured image 24 (step S160). Here, the second artificial neural network 46 recognizes the captured image 24 to obtain a second target image 26 of the selected target object 16 (step S170). The second target image 26 is a first target image 220 that overlaps with the mark images 240, 242 of the predetermined mark 52. Then, the comparison module 48 compares the second target image 26 with each previously stored first target image 220 (step S180), and the first target image 220 with the highest similarity to the second target image 26 is used as the selected image 480. The output module 70 extracts the string 450, 452 corresponding to the selected image 480 (i.e., the first target image 220 used as the selected image 480) from the storage module 60, and displays the string 450, 452 on the display 80 (step S190).

[0066] In this way, the user can quickly and conveniently select which barcode 122, 124 or string 146, 148 among the multiple barcodes 122, 124 or strings 146, 148 he or she wishes to output the string 450, 452 corresponding to the barcode 122, 124 or string 146, 148, and after the selection, quickly output the string 450, 452 corresponding to the selected barcode 122, 124 or string 146, 148 from the strings 450, 452 decoded or identified from the multiple barcodes 122, 124 or strings 148.

[0067] For example, if the pointing device 50 is an infrared pen or a laser pen, when the user presses the switch of the infrared pen or the laser pen, the infrared pen or the laser pen generates a cursor, and the marking image 240 of the cursor (i.e., the predetermined mark 52) begins to appear in the preset image 22 captured by the camera module 20 that continues to shoot. For example, if the finger is used as the pointing device 50, when the user's finger enters the photographing range of the camera module 20, the marking image 240 of the finger also begins to appear in the preset image 22 captured by the camera module 20.

[0068] refer to Figure 6B and Fig. 9In some embodiments, the processing module 40 includes a mark confirmation module 49, which is coupled to the camera module 20. The mark confirmation module 49 identifies the preset image 22 obtained by repeatedly shooting the camera module 20 to confirm whether the preset image 22 contains the mark images 240, 242 (step S150a), and after confirming that the preset image 22 contains the mark images 240, 242, the mark confirmation module 49 respectively identifies the position of the mark image 240 in the preset image 22 (step S150b), thereby knowing whether the mark image 240 overlaps with the first target image 220. When the mark image 240 overlaps with the first target image 220 and the overlapping time reaches a predetermined time (step S150c), the processing module 40 generates a confirmation signal and transmits it back to the camera module 20. After the camera module 20 detects the confirmation signal (step S150), it performs another single shooting to obtain the intercepted image 24 with the second target image 26 (step S160).

[0069] The purpose of generating the confirmation signal only after the images to be marked 240, 242 overlap with one of the first target images 220 for a predetermined time is that the predetermined mark 52 cannot accurately point to the specific target 10 at the beginning. For example, if the pointing device 50 is an infrared pen or a laser pen, the user often cannot directly place the cursor on the specific target 10 at the moment of pressing the switch, and needs to move the cursor from the position at the moment of pressing the switch to the specific target 10 (i.e., the selected target 16). In the embodiment where a finger is used as the pointing device 50, the same situation also occurs, and the finger cannot point to the specific target 10 at the moment it appears in the preset image 22.

[0070] During the movement of the cursor or finger, it may pass by other targets 10 besides the selected target 16, and the time that the finger overlaps with other targets 10 during the movement will be shorter. Therefore, whether the cursor or finger is actually on the selected target 16 can be judged by "whether it overlaps for a predetermined time" (step S150c).

[0071] In some embodiments, if the mark confirmation module 49 recognizes that the preset image 22 does not contain the mark image 240, 242 or that the preset image 22 contains the mark image 240, 242 but the time that the mark image 240, 242 overlaps with the first target image 220 does not exceed a predetermined time, it means that the preset mark 52 (finger or cursor) has not been accurately indicated on the selected target object 16, and there is no need to use a confirmation signal to make the camera module 20 take a single shot to obtain the captured image 24. At this time, the mark confirmation module 49 can repeatedly judge the preset image 22 and generate a confirmation signal when it meets the above-mentioned conditions that the preset image 22 contains the mark image 240, 242, the mark image 240, 242 overlaps with the first target image 220 and overlaps for a predetermined time.

[0072] Re-reference Figure 1B In some embodiments, the user generates a confirmation signal through an input element 30, and the input element 30 is a button 32. This button 32 can be directly implemented by a switch of an infrared pen or a laser pen. After pressing the button 32, a cursor appears on the infrared pen or the laser pen, and the camera module 20 also starts shooting. When the mark confirmation module 49 recognizes that the preset image 22 captured by the camera module 20 has a mark image 240, and the mark image 240 overlaps with the first target image 220 and overlaps for a predetermined time, a confirmation signal is generated. After detecting the confirmation signal, the camera module 20 generates a captured image 24 with a second target image 26.

[0073] In some embodiments, the input element 30 implemented by the button 32 is a button 32 independent of the switch of the infrared pen or the laser pen, and the user can press the button 32 when the finger is indeed overlapped with the selected target 16. The camera module 20 performs a single shot according to the confirmation signal generated when the user presses the button 32 to obtain the captured image 24 having the second target image 26.

[0074] refer to Figure 2 or Figure 7 , Figure 8 In some embodiments, after the selected image 480 is confirmed, the output module 70 extracts the character string 450, 452 stored after the decoder 440 decodes or the identifier 442 recognizes the text, and outputs the character string 450, 452 and displays it via the display 80. The display 80 can set a display field 82 for the character string 450, 452, and the character string 450, 452 is displayed in the display field 82 so that the user can intuitively obtain the information represented by the target object 10.

[0075] In some embodiments, the design of the columns in the display 80 can be adjusted according to user needs. For example, the display 80 can include the preset image 22 and the display column 82, the captured image 24 and the display column 82 (such as Figure 7 , Figure 8 ), or the second target image 26 is displayed in parallel with the selected image 480 and the display field 82 in the screen of the display 80. In some embodiments, the display field 82 can be presented in a window, which pops up after the processing module 40 determines that the selected image 480 is selected, and displays the string 450, 452 corresponding to the selected image 480.

[0076] refer to Fig. 6A , Figure 7 and Figure 8 , Fig.10In other embodiments, when the input element 30 is implemented as a button 32, the button 32 and the camera module 20 are located together on the portable electronic device 90. In some embodiments, the button 32 can be any button on the portable electronic device 90 or a button on the user interface of the portable electronic device 90, and the camera module 20 can be a front lens or a rear lens mounted on the portable electronic device 90.

[0077] In some embodiments, the camera module 20, the processing module 40, the storage module 60, the output module 70, and the display 80 may also be implemented by components of the portable electronic device 90 itself. For example, the processing module 40 and the output module 70 may be implemented by the processor 92 of the portable electronic device 90 itself, the internal memory in the processor 92 or the external memory of the portable electronic device 90 may serve as the storage module 60, and the display 80 may be implemented by the display screen of the portable electronic device 90. The indicating device 50 may also be disposed on the portable electronic device 90.

[0078] In other embodiments, the camera module 20, the processing module 40, the storage module 60, the output module 70, and the display 80 may also be independent of each other. For example, the camera module 20 is a camera that can be externally connected via a USB connection port or a Type C connection port, the processing module 40 is a processor 92 in the electronic device, the storage module 60 is an external hard disk, and the display 80 is an external screen.

[0079] In some other embodiments, the camera module 20 , the processing module 40 , the storage module 60 , the output module 70 , and the display 80 may be partially implemented by components of the portable electronic device 90 itself, and partially implemented by independent devices independent of the portable electronic device 90 .

[0080] In some embodiments, the portable electronic device 90 may be, for example, a smart phone, a notebook computer, a tablet computer, or the like.

[0081] In some embodiments, the processing module 40 may be implemented by one or more processors 92. Each processor may be, but is not limited to, a central processing unit, a system on chip (SOC), a general-purpose or special-purpose microprocessor, a digital signal processor (DSP), a programmable logic controller (PLC), an application specific integrated circuit (ASIC), a programmable logic device (PLD), other similar processing devices, or a combination of these devices.

[0082] In some embodiments, in addition to being implemented as the screen without a touch function of the portable electronic device 90, the display 80 may also be, but not limited to, a touch screen or a general external display screen (i.e., it has no touch function). Among them, the multiple preset images 22 captured in advance by the camera module 20 and captured in succession can be presented in real time on the display 80. When the display 80 is a touch screen, the input element 30 may be the display 80, or may be an input device such as a keyboard, one or more buttons, a handwriting board, or any combination thereof.

[0083] In some embodiments, the recognizer 442 recognizes the word strings 146, 148 by optical character recognition (OCR). In addition, the recognizer 442 can also be implemented by data annotation and supervised learning, template matching method, support vector machine (SVM) or decision tree method in machine learning model.

[0084] In some embodiments, the comparison module 48 uses a method of comparing the image similarity between the second target image 26 and each of the first target images 220 to obtain the selected image 480. The image similarity comparison method may be Mean Squared Error (MSE), Structural Similarity Index (SSIM), histogram comparison, feature point comparison, etc. Except for the portion in the second target image 26 that overlaps with the predetermined mark 52 or the finger, the other portions in the second target image 26 should be the same as one of the stored multiple first target images 220.

[0085] In some embodiments, the barcode image recognition device is used in warehouse or logistics management of goods. There are one or more barcodes 122, 124 or strings 146, 148 on the goods or shelves, wherein each barcode 122, 124 or string 146, 148 may carry relevant information such as product name, specification, origin, price and batch number. In some embodiments, these barcodes 122, 124 or strings 146, 148 may be set (such as pasted or printed) on the goods or shelves. During the processing, storage and transportation process, the barcode 122, 124 is directly scanned to obtain the information carried or the string 146, 148 is obtained.

[0086] In some embodiments, the barcodes 122 and 124 may have the same or different barcode types. Specifically, the barcodes 122 and 124 may have the same encoding method (i.e., the same barcode type) or different encoding methods (i.e., different barcode types). In some embodiments, the barcodes 122 and 124 may be, but are not limited to, any one-dimensional barcode (e.g., Code 39 barcode, Interleaved 25 barcode, EAN-13 barcode, EAN-8 barcode, or Code 128 barcode, etc.) or any two-dimensional barcode (e.g., QR Code barcode, PDF 417 barcode, composite barcode, or Data Matrix barcode, etc.). The one-dimensional barcode 122 and 124 may represent a string 450 of dozens of characters, and the two-dimensional barcode may also represent a string 450 of thousands of characters.

[0087] In some embodiments, the strings 146, 148 are not limited to languages, and the strings 146, 148 may represent the identification code of the goods to identify the specific goods, such as the Global Trade Item Number (GTIN, including UPC, EAN, etc.), the International Standard Book Number (ISBN, for books) or other related identification codes; the strings 146, 148 may also represent the manufacturer or supplier code (Manufacturer or Supplier Code), batch number (Batch Number) or serial number (Serial Number) that can be used to track the production batch or product serial number of the goods.

[0088] In summary, in some embodiments, the barcode image recognition method and device can accurately and controllably output the information carried by the indicated barcode 122, 124 or the converted string 146, 148 that can be displayed on the display 80 when there are multiple barcodes 122, 124 or strings 146, 148 at the same time.

[0089] Of course, the present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.

Claims

1. A barcode image recognition method, characterized in that: Include: Start a camera module; After the camera module is activated, the camera module is used to continuously pre-shoot at least one target object to obtain a plurality of preset images with a continuous shooting time; Using a first artificial neural network to recognize a first target image of each target object in each preset image; Extracting each of the first target images from each of the preset images; Convert each of the first target images into a string; storing each of the first target images and the character string corresponding to each of the first target images; Triggering the camera module to take a single shot of a given mark to obtain a captured image according to a confirmation signal; Using a second artificial neural network to identify a mark image of the predetermined mark in the captured image and a second target image of a selected target object among the at least one target object that overlaps with the predetermined mark; Comparing the second target image with each of the first target images to obtain a selected image, the selected image being the first target image having the highest similarity to the second target image; and The character string corresponding to the selected image is read from the stored character string and outputted.

2. The barcode image recognition method according to claim 1, characterized in that: Also includes: The first artificial neural network is used to identify a type of each of the first target images, wherein the type is one of a barcode image and a string image.

3. The barcode image recognition method according to claim 2, characterized in that: The first target image is the barcode image, and the step of converting each of the first target images into the character string includes: decoding the first target image into the character string, wherein the character string is information carried by the first target image.

4. The barcode image recognition method according to claim 2, characterized in that: The first target image is the character string image, and the step of converting each of the first target images into the character string includes: performing text recognition on the first target image to obtain the character string.

5. The barcode image recognition method according to claim 1, characterized in that: The method also includes receiving the confirmation signal from an input element.

6. The barcode image recognition method according to claim 5, characterized in that: The input element is a button.

7. The barcode image recognition method according to claim 6, characterized in that: The button is located on an indication device, and the indication device and the camera module are independent of each other.

8. The barcode image recognition method according to claim 7, characterized in that: Also includes: The indicating device is used to generate the predetermined mark on one of the at least one target object.

9. The barcode image recognition method according to claim 8, characterized in that: The indicating device is an infrared pen or a laser pen, and the predetermined mark is a cursor.

10. The barcode image recognition method according to claim 6, characterized in that: The button and the camera module are located on a portable electronic device.

11. The barcode image recognition method according to claim 10, characterized in that: Also includes: The portable electronic device is used to generate the predetermined mark on one of the at least one target object, wherein the predetermined mark is a cursor.

12. The barcode image recognition method according to claim 1, wherein: Also includes: Identifying the mark image of the predetermined mark in each of the preset images to obtain a mark position of the mark image in each of the preset images; and The confirmation signal is generated according to the shooting time and the mark position of the preset images having the mark image among the preset images.

13. The barcode image recognition method according to claim 12, characterized in that: Also includes: A pointing device is used to point to the selected target among the at least one target to form the predetermined mark overlapping with the selected target.

14. The barcode image recognition method according to claim 1, characterized in that: Also includes: The predetermined mark is provided to overlap the selected target among the at least one target.

15. A barcode image recognition device, characterized in that: Include: A camera module, used for continuously pre-shooting at least one target object to obtain a plurality of preset images with continuous shooting time; An input element, coupled to the camera module, for generating a confirmation signal to trigger the camera module to take a single shot of a predetermined mark to obtain a captured image in which the predetermined mark overlaps with the at least one target object; A processing module, coupled to the camera module, comprising: a first artificial neural network for identifying a first target image of each target object in each of the preset images; a conversion module, converting each of the first target images into a string; a second artificial neural network for identifying a marker image of the predetermined marker in the captured image and a second target image of a selected target object among the at least one target object that overlaps with the predetermined marker; and a comparison module, coupled to the first artificial neural network and the second artificial neural network, comparing the second target image with each of the first target images to obtain a selected image, the selected image being the first target image with the highest similarity to the second target image; a storage module, coupled to the processing module, and storing each of the first target images and the character string corresponding to each of the first target images; an output module coupled to the comparison module and the storage module, the output module reading the string corresponding to the selected image from the storage module; as well as A display is coupled to the output module, and the display displays the character string read by the output module.

16. The barcode image recognition device according to claim 15, characterized in that: The first artificial neural network also identifies a type of each of the first target images, which type is a barcode image or a string image.

17. The barcode image recognition device according to claim 16, characterized in that: The conversion module includes a decoder, and the decoder decodes the first target image into the string.

18. The barcode image recognition device according to claim 16, characterized in that: The conversion module includes a recognizer, which performs text recognition on the first target image to obtain the string.

19. The barcode image recognition device according to claim 15, characterized in that: The input element is a button.

20. The barcode image recognition device according to claim 19, characterized in that: The invention also comprises an indicating device, the button is located on the indicating device, and the indicating device and the camera module are independent of each other, and the indicating device generates the predetermined mark on one of the at least one target object.

21. The barcode image recognition device according to claim 20, characterized in that: The indicating device is an infrared pen or a laser pen, and the predetermined mark is a cursor.

22. The barcode image recognition device according to claim 19, characterized in that: The button and the camera module are located on a portable electronic device.

23. The barcode image recognition device according to claim 22, characterized in that: The portable electronic device generates the predetermined mark on one of the at least one target object.

24. The barcode image recognition device according to claim 15, characterized in that: The predetermined mark is a finger, a cursor generated by an infrared pen, or a cursor generated by a laser pen.

25. The barcode image recognition device according to claim 24, characterized in that: The processing module also includes a mark confirmation module for identifying the mark image of the predetermined mark in each of the preset images to obtain a mark position of the mark image in each of the preset images, and generating the confirmation signal according to the shooting time and the mark position of the multiple preset images having the mark image in the preset images.

Images