Method for acquiring vocabulary library used for recognizing visible and speaking functions of vehicle-mounted 3D map
Through the cloud OCR and post-processing technology of vehicle-computer interaction, the problem of incomplete scraping and error recognition of 3D map page elements of vehicle-computer 3D map is solved, and more accurate and fast voice-visible and can be used to improve the user experience.
Patent Information
- Application Number
- CN202410824706.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-25
- Publication Date
- 2024-11-22
AI Technical Summary
Traditional accessibility services cannot monitor the page elements of the 3D map of the car machine, resulting in incomplete page elements, incorrect recognition results, slow response, affecting user experience, and poor cloud configuration solutions, making map information impossible.
Through the cloud OCR of the vehicle-computer interaction, upload map images for text recognition, perform post-processing in the cloud, eliminate invalid information, correct error recognition results, and form a vocabulary library for the vehicle-computer 3D map that can be visible, and can be used for function recognition.
It improves the accuracy of OCR results, optimizes the response speed, supports the voice-visible and talkable function of the on-board 3D map, and improves the user experience.
Smart Images

Figure CN119003783A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of intelligent cockpit control, and in particular relates to a method for acquiring a vocabulary library used for identifying a visible-can-be-spoken function of a vehicle computer 3D map. Background Art
[0002] Visible and audible is an important task of the voice module in the smart cockpit. Visible and audible means that the current page elements can be controlled by voice on the car computer. For example, when opening QQ Music, saying "My Favorites" on the current page is equivalent to clicking the corresponding button to view the "My Favorites" function.
[0003] To achieve this effect, the main steps are as follows: 1. Speech recognition, converting audio into text intent through natural language understanding; 2. Capturing the current page element set; 3. Matching the intent with the current element set. Among them, the capture of the page element set, the technology commonly used by most car manufacturers at present is the barrier-free scanning service. The barrier-free service is a system service provided by systems such as Android and IOS. When a process starts the barrier-free service, it can monitor the changes of the foreground page elements and obtain all element node information (text, description) and index; after matching a specific node through user intent, it can initiate clicks, slides and other operations on the page element represented by the node according to the index of the node.
[0004] However, the accessibility service cannot monitor the car's 3D map. Since the map is 3D, it is impossible to monitor all text changes and descriptions on the foreground page, which will lead to incomplete capture of page elements. The large number of missing result sets will affect subsequent intent matching and greatly reduce the user experience.
[0005] In summary, traditional OCR technology is already very mature, but there are the following problems when using OCR algorithms to recognize car map images in the cloud:
[0006] 1. The recognition result is wrong. After investigation, it was found that the OCR recognition process is carried out by line, while a lot of information on the map is multiple lines of text gathered together to indicate a place name; for example, "Vanke Tomorrow Square South District" is two lines of information on the map, "Vanke Tomorrow" and "Square South District", so the place name will be recognized incorrectly, and the one place "Vanke Tomorrow Square South District" will be recognized as two places, "Vanke Tomorrow" and "Square South District", resulting in recognition errors.
[0007] 2. The response of "visible and speakable" is slow. The reason is that the result set after map recognition is too large, which takes a long time to match the result set with the voice registration and intent, affecting the overall "visible and speakable" link. Another reason is that the car map has a large amount of information, but the text information such as street names has no response when clicked, and this part of meaningless results has not been optimized.
[0008] In addition, for visible information, a cloud configuration solution can also be adopted, that is, all the text information on the map is configured and registered in the cloud. However, the effect of this solution is too poor and the limitations are too strong, because a large amount of information on the map cannot be exhaustively listed and cannot be fully configured by manpower. Summary of the invention
[0009] In order to overcome the above problems, the present invention provides a method for obtaining a vocabulary library used for vehicle-mounted 3D map visible and speakable function recognition, and performs page recognition through vehicle-mounted interactive cloud OCR. The vehicle-side uploads the current map image, and the cloud-based OCR algorithm performs recognition and returns the result to the vehicle-side. The OCR recognition result is post-processed by an improved OCR method, and the erroneous data is corrected and the meaningful results are screened and retained.
[0010] A method for obtaining a vocabulary library used for vehicle computer 3D map visible and audible function recognition includes the following contents:
[0011] Step 1: The vehicle captures the current desktop map interface image and uploads it to the cloud;
[0012] Step 2: The cloud performs text recognition on the image to obtain a set of place names in the map interface image;
[0013] Step 3: Perform the first step post-processing on the identified place name set to remove invalid information. The specific contents are as follows:
[0014] Step 3.1, traverse the place name set obtained in step 2;
[0015] Step 3.2, each time a word in the place name set is traversed, it is necessary to determine whether the word is in the set invalid word set. If so, the corresponding word is deleted from the place name set, otherwise it is retained in the place name set;
[0016] Step 3.3, end the traversal and obtain the final vocabulary set;
[0017] Step 4: Perform the second post-processing on the final vocabulary set to correct the incorrectly recognized place names. The specific contents are as follows:
[0018] Step 4.1, traverse the final vocabulary set;
[0019] Step 4.2, each time a word is traversed, calculate the distance d between the i-th place name corresponding to the current word and the i+1-th place name in the final word set; determine whether the distance d is less than the threshold; if it is less than, proceed to step 4.3; if not, set i=i+1 and repeat step 4.2;
[0020] Step 4.3, calculate the distance d between the i+1th place name and the i+2th place name i+1 , if the distance di+1 is greater than 15, go to step 4.4; if the distance d i+1 If it is not greater than 15, set i=i+1 and repeat step 4.3;
[0021] Step 4.4, concatenate the i-th, i+1-th, and i+2-th place name information to obtain the vocabulary used for the vehicle computer 3D map visible and audible function recognition, and delete the i+1-th and i+2-th place names in the final vocabulary set of step 4.1;
[0022] Step 4.5, let i=i+2 and repeat steps 4.2-4.5 until the final vocabulary set traversal is completed, and the vocabulary library used for the visible and audible function recognition of the vehicle 3D map is obtained;
[0023] Step 5. The vehicle side obtains the vocabulary library used for the visible-and-speak function recognition of the vehicle computer's 3D map and performs voice visible-and-speak processing.
[0024] The distance calculation formula between two place names in 4.2 is as follows:
[0025] D=(x2 _mid -x1 _mid ) 2 +(y2 _mid -y1 _mid ) 2
[0026] in:
[0027] x1_ mid =(x1_left_up+x1_right_down) / 2
[0028] y1 _mid= (y1_left_up+y1_right_down) / 2
[0029] x2 _mid =(x2_left_up+x2_right_down) / 2,
[0030] y2 _mid =(y2_left_up+y2_right_down) / 2
[0031] The coordinates of the four points of Place1 of the previous place name are: upper left (x1_left_up, y1_left_up), upper right (x1_right_up, y1_right_up), lower left (x1_left_down, y1_left_down), and lower right (x1_right_down, y1_right_down);
[0032] The coordinates of the four points of Place2 of the latter place name are: upper left (x2_left_up, y2_left_up), upper right (x2_right_up, y2_right_up), lower left (x2_left_down, y2_left_down), and lower right (x2_right_down, y2_right_down).
[0033] In step 2, the cloud calls the OCR algorithm to perform text recognition on the image.
[0034] The invalid information in step 3 is divided into two categories. The first category is the streets in the map, and the second category is custom words that can be spoken without the map voice being visible.
[0035] The street includes XX Street or XX Road.
[0036] The customized words that can be spoken without map voice visibility include charging identifiers.
[0037] The threshold in step 4.3 is set to 15.
[0038] Beneficial effects of the present invention:
[0039] According to the particularity of the in-vehicle 3D navigation map, the present invention provides an alternative solution for capturing the page element set of the in-vehicle 3D map, and in this solution, the accuracy of the OCR result is improved, the response result is optimized, and the voice visible and audible function of the in-vehicle 3D map is generally supported. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings used in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the contents of the embodiments of the present invention and these drawings without paying any creative work.
[0041] Figure 1 This is a flow chart of a part of the method of the present invention.
[0042] Figure 2 This is another part of the method flow chart of the present invention. DETAILED DESCRIPTION
[0043] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.
[0044] In the description of the present invention, unless otherwise clearly specified and limited, the terms "connected", "connected", and "fixed" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0045] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes that the first feature is directly below and obliquely below the second feature, or simply indicates that the first feature is lower in level than the second feature.
[0046] In the description of this embodiment, the terms "upper", "lower", "left", "right" and other directions or positional relationships are based on the directions or positional relationships shown in the drawings, and are only for the convenience of description and simplification of operation, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operate in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are only used to distinguish in the description and have no special meaning.
[0047] Example 1
[0048] A method for obtaining a vocabulary library used for vehicle computer 3D map visible and audible function recognition includes the following contents:
[0049] Step 1: The vehicle captures the current desktop map interface image and uploads it to the cloud;
[0050] Step 2: The cloud performs text recognition on the image to obtain a set of place names in the map interface image;
[0051] Step 3: Perform the first step post-processing on the identified place name set to remove invalid information. The specific contents are as follows:
[0052] Step 3.1, traverse the place name set obtained in step 2;
[0053] Step 3.2, each time a word in the place name set is traversed, it is necessary to determine whether the word is in the set invalid word set. If so, the corresponding word is deleted from the place name set, otherwise it is retained in the place name set;
[0054] Step 3.3, end the traversal and obtain the final vocabulary set;
[0055] Step 4: Perform the second post-processing on the final vocabulary set to correct the incorrectly recognized place names. The specific contents are as follows:
[0056] Step 4.1, traverse the final vocabulary set;
[0057] Step 4.2, each time a word is traversed, calculate the distance d between the i-th place name corresponding to the current word and the i+1-th place name in the final word set; determine whether the distance d is less than the threshold; if it is less than, proceed to step 4.3; if not, set i=i+1 and repeat step 4.2;
[0058] Step 4.3, calculate the distance d between the i+1th place name and the i+2th place name i+1 , if the distance d i+1 is greater than 15, go to step 4.4; if the distance d i+1 If it is not greater than 15, set i=i+1 and repeat step 4.3;
[0059] Step 4.4, concatenate the i-th, i+1-th, and i+2-th place name information to obtain the vocabulary used for the vehicle computer 3D map visible and audible function recognition, and delete the i+1-th and i+2-th place names in the final vocabulary set of step 4.1;
[0060] Step 4.5, let i=i+2 and repeat steps 4.2-4.5 until the final vocabulary set traversal is completed, and the vocabulary library used for the visible and audible function recognition of the vehicle 3D map is obtained;
[0061] Step 5. The vehicle side obtains the vocabulary library used for the visible-and-speak function recognition of the vehicle computer's 3D map and performs voice visible-and-speak processing.
[0062] The distance calculation formula between two place names in 4.2 is as follows:
[0063] D=(x2 _mid -x1 _mid ) 2 +(y2 _mid -y1 _mid ) 2
[0064] in:
[0065] x1_ mid=(x1_left_up+x1_right_down) / 2
[0066] y1 _mid= (y1_left_up+y1_right_down) / 2
[0067] x2 _mid =(x2_left_up+x2_right_down) / 2,
[0068] y2 _mid =(y2_left_up+y2_right_down) / 2
[0069] The coordinates of the four points of Place1 of the previous place name are: upper left (x1_left_up, y1_left_up), upper right (x1_right_up, y1_right_up), lower left (x1_left_down, y1_left_down), and lower right (x1_right_down, y1_right_down);
[0070] The coordinates of the four points of Place2 of the latter place name are: upper left (x2_left_up, y2_left_up), upper right (x2_right_up, y2_right_up), lower left (x2_left_down, y2_left_down), and lower right (x2_right_down, y2_right_down).
[0071] In step 2, the cloud calls the OCR algorithm to perform text recognition on the image.
[0072] The invalid information in step 3 is divided into two categories. The first category is the streets in the map, and the second category is custom words that can be spoken without the map voice being visible.
[0073] The street includes XX Street or XX Road.
[0074] The customized words that can be spoken without map voice visibility include charging identifiers.
[0075] The threshold in step 4.3 is set to 15.
[0076] Example 2
[0077] 1. The vehicle captures the current desktop map and uploads it to the cloud.
[0078] 2. The cloud uses the OCR algorithm to perform text recognition on the image and obtain the set of place names in the map.
[0079] 3. Perform the first step of post-processing on the recognition vocabulary set to eliminate invalid information.
[0080] 3.1, traverse the vocabulary set.
[0081] 3.2, Is the word in the invalid word set? If yes, delete it, otherwise keep it.
[0082] 3.3, Is the word "XX Street", for example, "Qingnian Street"? If yes, delete it, otherwise keep it.
[0083] 3.4, end the traversal and obtain the final vocabulary set.
[0084] Note: Invalid information is divided into two categories: one is the streets on the map, such as "XX Street", "XX Road"; the other is a set of custom words that can be spoken without the need for map voice visibility, such as the "toll" symbol on the map. Figure 1 .
[0085] 4. Perform the second step of post-processing on the recognition vocabulary set to correct the incorrect recognition algorithm of place names, such as Figure 2 .
[0086] 4.2, traverse the result set and get the current place name.
[0087] 4.3, calculate the distance between the current place name i and the next place name i+1.
[0088] 4.4, determine whether the distance is less than the threshold? (The threshold is set by yourself, and it is 15 according to practice)
[0089] 4.5, if yes, then continue to calculate the distance between the i+1th and i+2th place names until the distance is greater than 15
[0090] 4.6 Concatenate the above-mentioned i-th, i+1-th, and i+2-th place names, and delete the i+1 and i+2 place names.
[0091] 4.7, now i=i+2 and continue to repeat (2), (3), (4), and (5) until the set traversal is completed.
[0092] The formula for calculating the distance between two consecutive place names is described as follows:
[0093] (1) The coordinates of the four points of Place1: upper left (x1_left_up, y1_left_up), upper right (x1_right_up, y1_right_up), lower left (x1_left_down, y1_left_down), and lower right (x1_right_down, y1_right_down).
[0094] (2) Coordinates of the four points of Place2: upper left (x2_left_up, y2_left_up), upper right (x2_right_up, y2_right_up), lower left (x2_left_down, y2_left_down), and lower right (x2_right_down, y2_right_down).
[0095] (3) Midpoint coordinates:
[0096] Place 1_mid (x1 _mid , y1 _mid ), Place 2_mid (x2 _mid , y2 _mid )
[0097] x1_ mid =(x1_left_up+x1_right_down) / 2,
[0098] y1 _mid= (y1_left_up+y1_right_down) / 2
[0099] x2 _mid =(x2_left_up+x2_right_down) / 2,
[0100] y2 _mid =(y2_left_up+y2_right_down) / 2
[0101] (4) Midpoint distance D = (x2 _mid -x1 _mid ) 2 +(y2 _mid -y1 _mid ) 2
[0102] 5. After the vehicle obtains the recognition result set, it performs subsequent voice visible and audible processing, which is irrelevant to the present invention and will not be elaborated on.
[0103] The preferred embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the protection scope of the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention. These simple variations all belong to the protection scope of the present invention.
[0104] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present invention will not further describe various possible combinations.
[0105] In addition, various embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the present invention, they should also be regarded as the contents disclosed by the present invention.
Claims
1. A method for obtaining a vocabulary library used for identifying the visible and audible function of a vehicle-mounted 3D map, It is characterized in that It includes the following: Step 1: The vehicle captures the current desktop map interface image and uploads it to the cloud; Step 2: The cloud performs text recognition on the image to obtain a set of place names in the map interface image; Step 3: Perform the first step of post-processing on the identified place name set to remove invalid information. The specific contents are as follows: Step 3.1, traverse the place name set obtained in step 2; Step 3.2, each time a word in the place name set is traversed, it is necessary to determine whether the word is in the set invalid word set. If so, the corresponding word is deleted from the place name set, otherwise it is retained in the place name set; Step 3.3, end the traversal and obtain the final vocabulary set; Step 4: Perform the second post-processing on the final vocabulary set to correct the incorrectly recognized place names. The specific contents are as follows: Step 4.1, traverse the final vocabulary set; Step 4.2, each time a word is traversed, the distance d between the i-th place name corresponding to the current word and the i+1-th place name in the final word set is calculated; and whether the distance d is less than a threshold is determined; If it is less than, go to step 4.
3. If it is not less than, set i=i+1 and repeat step 4.
2. Step 4.3, calculate the distance d between the i+1th place name and the i+2th place name i+1 , if the distance d i+1 is greater than 15, go to step 4.4; if the distance d i+1 If it is not greater than 15, set i=i+1 and repeat step 4.3; Step 4.4, concatenate the i-th, i+1-th, and i+2-th place name information to obtain the vocabulary used for the vehicle computer 3D map visible and audible function recognition, and delete the i+1-th and i+2-th place names in the final vocabulary set of step 4.1; Step 4.5, let i=i+2 and repeat steps 4.2-4.5 until the final vocabulary set traversal is completed, and the vocabulary library used for the visible and audible function recognition of the vehicle 3D map is obtained; Step 5. The vehicle side obtains the vocabulary library used for the visible-and-speak function recognition of the vehicle computer's 3D map and performs voice visible-and-speak processing.
2. A method for acquiring a vocabulary library used for identifying a visible and audible function of a vehicle-mounted 3D map according to claim 1, It is characterized in that The distance calculation formula between two place names in 4.2 is as follows: D=(x2 _mid -x1 _mid ) 2 +(y2 _mid -y1 _mid ) 2 in: x1_ mid =(x1_left_up+x1_right_down) / 2 y1 _mid= (y1_left_up+y1_right_down) / 2 x2 _mid =(x2_left_up+x2_right_down) / 2, y2 _mid =(y2_left_up+y2_right_down) / 2 The first place name is Place 1 The coordinates of the four points are: upper left (x1_left_up, y1_left_up), upper right (x1_right_up, y1_right_up), lower left (x1_left_down, y1_left_down), lower right (x1_right_down, y1_right_down); Place 2 The coordinates of the four points are: upper left (x2_left_up, y2_left_up), upper right (x2_right_up, y2_right_up), lower left (x2_left_down, y2_left_down), and lower right (x2_right_down, y2_right_down).
3. The method for acquiring a vocabulary library used for identifying a visible and audible function of a vehicle-mounted 3D map according to claim 1, It is characterized in that In step 2, the cloud calls the OCR algorithm to perform text recognition on the image.
4. The method for acquiring a vocabulary library used for identifying a visible and audible function of a vehicle-mounted 3D map according to claim 1, It is characterized in that The invalid information in step 3 is divided into two categories. The first category is the streets in the map, and the second category is custom words that can be spoken without the map voice being visible.
5. A method for acquiring a vocabulary library used for identifying a 3D map visible and audible function of a vehicle computer according to claim 4, It is characterized in that The street includes XX Street or XX Road.
6. A method for acquiring a vocabulary library used for identifying a vehicle-mounted 3D map visible and audible function according to claim 1, It is characterized in that The customized words that can be spoken without map voice visibility include charging identifiers.
7. The method for acquiring a vocabulary library used for identifying a visible and audible function of a vehicle-mounted 3D map according to claim 1, It is characterized in that The threshold in step 4.3 is set to 15.
8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the program, the method for acquiring the vocabulary library used for the visible and audible function recognition of the vehicle machine 3D map as described in any one of claims 1-7 is implemented.
9. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the program is executed by a processor, the method for acquiring a vocabulary library used for identifying a visible and audible function of a vehicle-mounted 3D map as described in any one of claims 1 to 7 is implemented.
