Corner point detection method and apparatus

Abstract

The application provides a corner point detection method and device, and the method comprises the following steps: receiving a candidate corner point, and adding index information to the candidate corner point; sorting the candidate corner point with index information to determine the position of the candidate corner point with index information in a corner point sequence; and according to a set suppression distance, suppressing a target corner point in the corner point sequence to remove the target corner point in the corner point sequence. The application only needs to modify the index of the corner point to calibrate the sorting relationship of the corner point, does not need to adjust the position relationship of the corner point, is convenient for hardware implementation, can greatly reduce the operation frequency of the cache, and significantly reduces the operation time consumption; the application opens a small space cache in the hardware, and all operation operations are completed in the cache buffer, so that the detection effect is ensured, the frequent operation on the DDR is avoided, and the occupation of the DDR bandwidth is reduced.

Description

Technical Field

[0001] This application generally relates to the field of image processing technology, and in particular to corner detection methods and apparatus. Background Technology

[0002] Corner detection is a method for obtaining image features in computer vision systems. Due to its real-time performance and stability, corner detection is widely used in motion detection, image matching, video tracking, 3D modeling, and object recognition. Corners are considered feature points, hence corner detection is also called feature point detection. Corner detection typically uses the STcorner method, or Shi-Tomasi method. The Shi-Tomasi method is an improvement on the traditional Harris method, achieving better results. However, as a gradient-based method, it has high computational complexity. Especially after calculating qualified candidate corners, it is necessary to sort the candidate corners in descending order of quality and suppress other corners within the minimum distance range of high-quality corners. After multiple iterations, the optimal corner sequence with a specified number of points is finally returned.

[0003] In practical applications, the number of candidate corner points is enormous, and the sorting operation consumes a significant amount of cache space. Furthermore, multiple iterations are required to suppress other corner points within the minimum distance range of high-quality corner points, resulting in a massive computational load and long processing time. Current algorithms utilize DDR (Double Data Rate) for data caching, first sorting the corner points and then repeatedly suppressing the data in DDR within the minimum distance range. However, this operation consumes DDR bandwidth for extended periods, is inefficient, slow, and unsuitable for hardware implementation. Summary of the Invention

[0004] The embodiments of this disclosure provide a corner detection method and apparatus, which can be used to improve corner detection efficiency and reduce DDR bandwidth usage.

[0005] In a first aspect, a corner detection method is provided. The method includes: receiving candidate corners and adding index information to the candidate corners to generate candidate corners with index information; sorting the candidate corners with index information to determine the positions of the candidate corners with index information in a corner sequence; and suppressing target corners in the corner sequence according to a set suppression distance to remove the target corners from the corner sequence.

[0006] In some embodiments, the index information added to the candidate corner point includes: the position information of the previous corner point and the position information of the next corner point.

[0007] In some embodiments, the method further includes: constructing a lookup table, wherein the lookup table includes index information for a plurality of preset corner points.

[0008] In some embodiments, when a candidate corner point is received, the index information of the candidate corner point is iteratively narrowed down by comparing it with the index information of each preset corner point.

[0009] In some embodiments, sorting candidate corner points with index information includes: sorting the candidate corner points with index information based on the corner point sequence within the search range.

[0010] In some embodiments, sorting candidate corner points with index information further includes: after sorting the candidate corner points, storing the index information of the candidate corner points in the lookup table.

[0011] In some embodiments, sorting the candidate corner points with index information includes: sequentially comparing the information of the candidate corner point to be processed with the index information of each corner point in the corner point sequence stored in the cache, and determining the position of the candidate corner point with index information in the corner point sequence includes: modifying the index information of the candidate corner point to be processed according to the position of the candidate corner point to be processed in the corner point sequence.

[0012] In some embodiments, the step of suppressing the target corner in the corner sequence according to a set suppression distance to remove the target corner from the corner sequence includes: iteratively executing in an internal caching module: sequentially selecting corners in the sorted corner sequence in descending order as the suppression benchmark, filtering the corners sorted after them; and when the distance between a subsequent corner in the corner sequence and the current corner used as the suppression benchmark is less than the set suppression distance, filtering out the subsequent corner from the corner sequence.

[0013] In a second aspect, a corner detection apparatus is provided. The apparatus includes: an index information adding module configured to receive candidate corners and add index information to the candidate corners to generate candidate corners with index information; a sorting module configured to sort the candidate corners with index information to determine the positions of the candidate corners with index information in a corner sequence to generate a sorted corner sequence; and a corner suppression module configured to suppress target corners in the corner sequence according to a set suppression distance to remove the target corners from the corner sequence to generate an updated corner sequence.

[0014] In some embodiments, the system further includes a lookup table module, configured to construct a lookup table including index information of several preset corner points, and to iteratively narrow the search range by sequentially comparing the index information of the candidate corner points with the index information of each preset corner point.

[0015] In some embodiments, the system further includes a caching module configured to cache index information of each corner point in the corner point sequence, the sorted corner point sequence, and the updated corner point sequence.

[0016] In some embodiments, the corner suppression module is further configured to output the updated corner sequence to a memory external to the corner detection device.

[0017] In the corner detection method and apparatus according to embodiments of this disclosure, only the corner index needs to be modified to determine the sorting relationship of the corners, without needing to adjust the positional relationship of the corners. This facilitates hardware implementation and significantly reduces computation time. Furthermore, by allocating a small buffer within the hardware, all computational operations are completed within this buffer. This method ensures detection effectiveness while avoiding frequent DDR operations, thereby reducing DDR bandwidth usage. Attached Figure Description

[0018] Figure 1 The flowchart shown is a corner detection method according to an embodiment of the present disclosure;

[0019] Figure 2 The diagram shown is a schematic representation of a corner index in a corner detection method according to an embodiment of the present disclosure;

[0020] Figure 3 The diagram shows a cache storage structure in a corner detection method according to an embodiment of the present disclosure.

[0021] Figure 4 The diagram shows a lookup table in a corner detection method according to an embodiment of the present disclosure;

[0022] Figure 5 The diagram illustrates a method for narrowing the sorting range using a lookup table in a corner detection method according to an embodiment of this disclosure.

[0023] Figure 6 The diagram shows an update of the lookup table in a corner detection method according to an embodiment of the present disclosure;

[0024] Figures 7 to 10 The diagram shows a corner sorting process in a corner detection method according to an embodiment of the present disclosure.

[0025] Figure 11 The diagram shows the principle of corner suppression in a corner detection method according to an embodiment of the present disclosure;

[0026] Figure 12 The diagram shows the detection process principle of the corner detection method according to the embodiments of this disclosure;

[0027] Figure 13 The diagram shown is a schematic representation of a corner detection device according to an embodiment of the present disclosure. Detailed Implementation

[0028] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0029] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. Therefore, the drawings only show the components related to this application and are not drawn according to the number, shape and size of the components in actual implementation. In actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0030] In the aforementioned existing corner detection schemes, DDR bandwidth is occupied for a long time, and the efficiency and speed are low, making them unsuitable for hardware implementation.

[0031] To address at least the aforementioned technical problems, this disclosure provides a corner detection method and apparatus, which can be used in a hardware-based design architecture to achieve corner detection, significantly reducing computation time and DDR bandwidth usage while maintaining the original performance.

[0032] To address the issues of long processing time and high DDR bandwidth consumption in existing corner detection algorithms, embodiments of this disclosure establish an additional index structure for corner points. This index structure indicates the sorting relationship of the corner points, and only the index needs to be modified to determine the sorting relationship, thus eliminating the need to adjust the positional relationships of the corner points. In this way, the corner detection scheme according to embodiments of this disclosure is easy to implement in hardware, significantly reduces the frequency of cache operations, and substantially reduces computation time.

[0033] Furthermore, in the embodiments according to this disclosure, based on the characteristic of corner minimum distance range suppression, the sorting operation can be divided into several segments, thereby further reducing the required cache space capacity. In addition, by allocating a small cache space within the hardware, all computational operations are completed within this cache, avoiding frequent DDR operations and thus reducing DDR bandwidth usage.

[0034] Hereinafter, specific embodiments of the corner detection method and apparatus according to the present disclosure will be described with reference to the accompanying drawings and exemplary embodiments.

[0035] Figure 1 This is a flowchart illustrating a corner detection method 100 according to an embodiment of the present disclosure. Figure 1 As shown, the corner detection method 100 includes the following steps S102 to S106.

[0036] In step S102, candidate corner points are received and index information is added to the candidate corner points to generate candidate corner points with index information.

[0037] In some embodiments, the index information includes the position information of the preceding and following corner points of the candidate corner point. In this way, the sequential relationship of the candidate corner point quality sequence is established using the index information.

[0038] In step S104, the candidate corner points with index information are sorted to determine the position of the candidate corner points with index information in the corner point sequence.

[0039] In some embodiments, when sorting candidate corner points with index information, the candidate corner points with index information are sorted based on the corner point sequence within the search range determined by a lookup table to generate a sorted corner point sequence. In some embodiments, the lookup table includes index information of several preset corner points, and the search range is iteratively narrowed by sequentially comparing the index information of the candidate corner points with the index information of each preset corner point.

[0040] In step S106, the target corner points in the corner point sequence are suppressed according to the set suppression distance to remove the target corner points from the corner point sequence.

[0041] In some embodiments, corner points are selected sequentially from largest to smallest in the sorted corner point sequence as suppression benchmarks, and subsequent corner points are filtered. When the distance between a subsequent corner point in the corner point sequence and the corner point used as the suppression benchmark is less than the set suppression distance, the subsequent corner point is removed from the corner point sequence to update the corner point sequence.

[0042] In embodiments according to this disclosure, by establishing index information, the sorting relationship of corner points can be determined simply by modifying the index information, thus eliminating the need to adjust the positional relationships of the corner points. Furthermore, in embodiments according to this disclosure, a cache is set up to perform the operations associated with sorting and suppression within the cache, and the updated corner point sequence is sent to the DDR memory for storage. In this way, frequent operations on the DDR are avoided, thereby reducing the DDR bandwidth usage.

[0043] The following combination Figures 2 to 11 The corner detection method of this embodiment will be described in detail.

[0044] Figure 2 This diagram illustrates the corner index in the corner detection method of this embodiment. For example... Figure 2 As shown, in this embodiment, the index information added to the candidate corner point includes: the position information of the previous corner point and the position information of the next corner point.

[0045] This embodiment provides information about the current corner point's position within the ordered corner point sequence. Subsequent corner points can determine their position in the sequence based on the index information, and the entire sequence's order is updated by modifying the index. This allows the corner detection method in this embodiment to determine the corner point's order by simply modifying the corner point's index, without needing to adjust the corner points' positions. This simplifies hardware implementation, significantly reduces the frequency of cache operations, and substantially reduces computation time.

[0046] In this embodiment, candidate corner points with added index information are stored in a buffer. Figure 3 This diagram illustrates the cache storage structure in the corner detection method of this embodiment. (See attached diagram.) Figure 3 As shown, the buffer stores corner information and buffer identifiers at each space address (address n, address n+1, address n+2, ..., address m). In addition to storing candidate corners after adding index information, the buffer also stores intermediate temporary data during the corner detection process, enabling frequent interactions during internal hardware operations.

[0047] In this embodiment, the method further includes: constructing a lookup table, wherein the lookup table stores index information of several preset corner points (key corner points). Figure 4 This diagram illustrates the lookup table used in the corner detection method of this embodiment. For example... Figure 4As shown, the lookup table is divided into several table units: table unit 0, table unit 1, table unit 2, table unit 3, table unit 4, table unit 5, table unit 6, ..., table unit n. Each table unit stores the index information of a preset corner point. The index information of the preset corner point includes an index identifier, preset corner point information, and pointing position information.

[0048] The preset corner information includes the maximum corner information, the minimum corner information, and one corner information from the corner information that divides the corner sequence into several equal parts. For example, if the corner sequence is divided into 16 equal parts, the preset corner information includes the maximum corner information, the minimum corner information, the corner information at 1 / 16 position, the corner information at 1 / 8 position, the corner information at 3 / 16 position, the corner information at 1 / 4 position, the corner information at 5 / 16 position, ..., the corner information at 15 / 16 position.

[0049] In this embodiment, when a candidate corner point is received, the search range is iteratively narrowed by comparing the candidate corner point with the index information of each preset corner point, thereby narrowing the scope of the sorting operation. Then, the processed corner point is compared sequentially with the corner points within the search range in the buffer to determine the position of the currently processed corner point in the sequence, and the index information is modified accordingly.

[0050] Figure 5 This diagram illustrates how a lookup table is used to narrow down the sorting range in the corner detection method of this embodiment. For example... Figure 5 As shown, for example, if the corner to be processed is corner k, a query operation is initiated to the lookup table. Based on the index information of corner k, the search starts from the corner at position 1 / 2 in the lookup table. The index information of corner k is directly compared with the index information of the corner at position 1 / 2. If the index information of corner k is greater than the index information of the corner at position 1 / 2, the search continues to the range after the corner at position 1 / 2 in the lookup table. If the index information of corner k is greater than the index information of the corner at position 1 / 2, the search continues to the range before the corner at position 1 / 2 in the lookup table, thus narrowing the search range. The above process is repeated to continuously narrow the search range. For example, if the index information of corner k is found in the range starting from corner 15 / 16, the corresponding corner position information is obtained from the buffer to reduce the amount of computation required for sorting.

[0051] Therefore, in this embodiment, each time a new candidate corner point is processed, the index information of the candidate corner point is directly compared with the index information of the preset corner point (key corner point) to quickly find the range of the corner point, reduce the amount of computation required for sorting, and improve the search speed.

[0052] Furthermore, in this embodiment, sorting the candidate corner points with index information further includes: after the candidate corner points are sorted, storing the index information of the candidate corner points in the lookup table. That is, when the candidate corner points are sorted, the corner point may be updated in the lookup table.

[0053] In this embodiment, the construction of the lookup table is an iterative process. The lookup table is gradually built up based on existing candidate corner points and is continuously updated. When a new candidate corner point is processed, a query operation is initiated to the lookup table, and the lookup table returns the query result. When a candidate corner point is processed, because the positional relationship between the corner point sequences has changed, the lookup table needs to be updated and reconstructed accordingly based on the positional relationship of the candidate corner points. Figure 6 This diagram illustrates the updating of the lookup table in the corner detection method of this embodiment. For example... Figure 6 As shown, after finding the sorting position of the candidate corner point, the index information of the corner points before and after the candidate corner point in the buffer is updated. Then, this information is synchronously updated back to the lookup table, that is, the modification is completed and the entire lookup table is updated.

[0054] Figures 7 to 10 This diagram illustrates the corner sorting process in the corner detection method of this embodiment. In this embodiment, as... Figures 7 to 10 As shown, sorting the candidate corner points with index information includes: sorting the candidate corner points with index information based on the corner point sequence within the search range.

[0055] Specifically, comparing the information of the candidate corner point to be processed with the information of each corner point in the corner point sequence stored in the cache in turn, and determining the position of the candidate corner point with index information in the corner point sequence includes: modifying the index information of the candidate corner point to be processed according to the position of the candidate corner point to be processed in the corner point sequence.

[0056] like Figure 7 As shown, the index information of the corner point k to be sorted is compared with the index information of corner points a and x in the buffer. If the index information of the corner point k to be sorted is greater than the index information of corner point x, then the position of the next corner point y is obtained from corner point x, as shown. Figure 8 As shown, the index information of the corner point k to be sorted is further compared with the index information of the corner point y. If the index information of the corner point k to be sorted is less than the index information of the corner point y, then it is determined that the corner point k to be sorted should be located between the positions of the corner points x and y. Figure 9 As shown, determine the storage location of the corner point k to be sorted in the buffer, as follows: Figure 10 As shown, the index information of the corner point x and the index information of the corner point y before and after the corner point k to be sorted are modified at the same time.

[0057] Therefore, the corner detection method in this embodiment only needs to modify the corner index to determine the sorting relationship of the corners, without adjusting the positional relationship of the corners. This facilitates hardware implementation, greatly reduces the frequency of cache operations, and significantly reduces computation time. Moreover, by allocating a small cache space within the hardware, all computational operations are completed within this cache, ensuring detection effectiveness while avoiding frequent DDR operations, thereby reducing DDR bandwidth usage.

[0058] Figure 11 This diagram illustrates the principle of corner suppression in a corner detection method according to an embodiment of this application. In this embodiment, suppressing target corners in the corner sequence according to a set suppression distance to remove the target corners from the corner sequence includes: iteratively executing in a cache or memory: sequentially selecting corners in the sorted corner sequence in descending order as suppression benchmarks, and filtering points sorted after them; and when the distance between a subsequent corner in the corner sequence and the current corner used as the suppression benchmark is less than the set suppression distance, filtering out the subsequent corner from the corner sequence.

[0059] like Figure 11 As shown, assuming the minimum suppression distance is d, the suppression of corner points in the corner point sequence is performed iteratively in the buffer according to the order of the corner point sequence, following this process: First, the current corner point to be processed is used as the suppression corner point, and subsequent corner points are suppressed. If there are other corner points within the minimum suppression distance d of the current corner point to be processed, then corner points within these ranges are screened. That is, high-quality corner points are selected sequentially from the corner point sequence, all corner points affected by them in the sequence are suppressed, and the process is iterated in the buffer.

[0060] Figure 12 This is a flowchart illustrating the detection process principle of the corner detection method in this embodiment. Figure 12As shown, upon receiving input candidate corner points, index information is first added to the candidate corner points. This index information establishes the sequential relationship of the candidate corner point quality sequence, and the candidate corner points with added index information are stored in a buffer. Then, upon receiving a candidate corner point, the index information of the candidate corner point is iteratively compared with the index information of each preset corner point to narrow the search range, thereby reducing the scope of the sorting operation. Next, the candidate corner points with index information are sorted: the information of the candidate corner point to be processed is compared sequentially with the information of each corner point in the corner point sequence stored in the buffer to determine the position of the candidate corner point to be processed in the corner point sequence. Based on the position of the candidate corner point to be processed in the corner point sequence, the index information of the candidate corner point to be processed is modified. Therefore, the corner point detection method in this embodiment only needs to modify the corner point index to determine the sorting relationship of the corner points, without needing to adjust the positional relationship of the corner points. Finally, iterative execution is performed in the cache: in the sorted corner point sequence, corner points are selected in descending order as the basis for suppression, and subsequent corner points in the corner point sequence are suppressed: when the distance between the subsequent corner point in the corner point sequence and the current corner point to be processed is less than the set suppression distance, the subsequent corner point is filtered out from the corner point sequence.

[0061] like Figure 13 As shown, this embodiment also provides a corner detection device 100. The corner detection device 100 includes at least an index information adding module 110, a sorting module 120, and a corner suppression module 130.

[0062] In this embodiment, the index information adding module 110 is configured to receive candidate corner points and add index information to the candidate corner points to generate candidate corner points with index information.

[0063] Specifically, in this embodiment, index information is added to the candidate corner points, and the sequential relationship of the candidate corner point quality sequence is established through the index information.

[0064] This embodiment provides information about the current corner point's position within the ordered corner point sequence. Subsequent corner points can determine their position in the sequence based on the index information, and the entire sequence's order is updated by modifying the index. This allows the corner detection method in this embodiment to determine the corner point's order by simply modifying the corner point's index, without needing to adjust the corner points' positions. This simplifies hardware implementation, significantly reduces the frequency of cache operations, and substantially reduces computation time.

[0065] In this embodiment, the sorting module 120 is configured to sort the candidate corner points with indexed information to determine the position of the candidate corner points with indexed information in the corner point sequence, so as to generate a sorted corner point sequence. Sorting the candidate corner points with indexed information includes: sorting the candidate corner points with indexed information based on the corner point sequence within the search range.

[0066] In this embodiment, sorting the candidate corner points with index information includes: sequentially comparing the information of the candidate corner point to be processed with the information of each corner point in the corner point sequence stored in the cache, and determining the position of the candidate corner point with index information in the corner point sequence includes: modifying the index information of the candidate corner point to be processed according to the position of the candidate corner point to be processed in the corner point sequence.

[0067] In this embodiment, the corner detection device 100 may further include a lookup table module 150. The lookup table module 150 is configured to construct a lookup table including index information of several preset corners, and to iteratively narrow the search range by sequentially comparing the index information of the candidate corners with the index information of each preset corner.

[0068] Therefore, in this embodiment, each time a new candidate corner point is processed, the index information of the candidate corner point is directly compared with the index information of the preset corner point (key corner point) to quickly find the range of the corner point, reduce the amount of computation required for sorting, and improve the search speed.

[0069] In this embodiment, the corner suppression module 130 is configured to suppress target corners in the corner sequence according to a set suppression distance to remove the target corners from the corner sequence, thereby generating an updated corner sequence. The corner suppression module 130 stores the result in a memory module (DDR, Double Data Rate) 200.

[0070] In this embodiment, the corner detection device 100 may further include a cache module 140. The cache module 140 is configured to cache the index information of each corner in the corner sequence, the sorted corner sequence, and the updated corner sequence.

[0071] Specifically, in this embodiment, the corner suppression module 130 is further configured to output the updated corner sequence to a memory outside the corner detection device.

[0072] The corner suppression module 130 suppresses target corners in the corner sequence according to a set suppression distance to remove the target corners from the corner sequence, including: iteratively executing in the cache module 140 or the memory module 200: using the current corner to be processed as the suppression corner, suppressing subsequent corners in the corner sequence: when the distance between the subsequent corner in the corner sequence and the current corner to be processed is less than the set suppression distance, the subsequent corner is filtered out from the corner sequence.

[0073] The principle and implementation of the corner detection device 100 in this embodiment are the same as or similar to the principle and implementation of the corner detection method in Embodiment 1. The parts that are the same between the embodiments will not be described again.

[0074] In summary, this invention only requires modifying the corner point indices to determine the corner point sorting relationship, without needing to adjust the corner point positions, thus simplifying hardware implementation and significantly reducing computation time. Furthermore, by allocating a small cache space within the hardware, all computational operations are completed within this cache buffer, ensuring detection effectiveness while avoiding frequent DDR operations, thereby reducing DDR bandwidth usage. Therefore, this application effectively overcomes the various shortcomings of existing technologies and possesses high industrial applicability.

[0075] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A corner detection method, characterized in that, include: Receive candidate corner points and add index information to the candidate corner points, the index information being used to indicate the sorting relationship of the corner points; Construct a lookup table, which includes index information for several preset corner points. The index information for the preset corner points includes an index identifier, preset corner point information, and pointing position information. The search range is iteratively narrowed by sequentially comparing the index information of the candidate corner points with the index information of each preset corner point. Based on the corner point sequence within the search range, the candidate corner points with index information are sorted to determine the position of the candidate corner points with index information in the corner point sequence; as well as The target corner points in the corner point sequence are suppressed according to the set suppression distance, so as to remove the target corner points from the corner point sequence; Sort the candidate corner points with index information by: sequentially comparing the index information of the candidate corner points to be processed with the index information of each corner point in the corner point sequence stored in the cache, and Determining the position of the candidate corner point with index information in the corner point sequence includes: modifying the index information of the candidate corner point to be processed according to its position in the corner point sequence.

2. The corner detection method according to claim 1, characterized in that, Adding index information to the candidate corner point includes adding the position information of the previous corner point and the position information of the next corner point.

3. The corner detection method according to claim 1, characterized in that, The sorting of candidate corner points with index information also includes: after sorting the candidate corner points, storing the index information of the candidate corner points in the lookup table.

4. The corner detection method according to claim 1, characterized in that, Suppressing target corners in the corner sequence according to a set suppression distance to remove the target corners from the corner sequence includes: Iterative execution within the cache: In the sorted corner point sequence, corner points are selected sequentially from largest to smallest as the suppression benchmark, and subsequent corner points are filtered out; and... When the distance between a subsequent corner point in the corner point sequence and the corner point used as the suppression reference is less than the set suppression distance, the subsequent corner point is filtered out from the corner point sequence.

5. A corner detection device, characterized in that, include: The index information adding module is configured to receive candidate corner points and add index information to the candidate corner points to generate candidate corner points with index information, wherein the index information is used to indicate the sorting relationship of the corner points; The lookup table module is configured to construct a lookup table including index information of several preset corner points, and to iteratively narrow the search range by sequentially comparing the index information of the candidate corner points with the index information of each preset corner point. The index information of the preset corner points includes an index identifier, preset corner point information, and pointing position information. The sorting module is configured to sort the candidate corners with index information based on the corner sequence within the search range to determine the position of the candidate corners with index information in the corner sequence, so as to generate a sorted corner sequence. Sort the candidate corner points with index information, including: sequentially comparing the index information of the candidate corner point to be processed with the index information of each corner point in the corner point sequence stored in the cache; and determining the position of the candidate corner point with index information in the corner point sequence, including: modifying the index information of the candidate corner point to be processed according to its position in the corner point sequence; and A corner suppression module is configured to suppress target corners in the corner sequence according to a set suppression distance to remove the target corners from the corner sequence, thereby generating an updated corner sequence.

6. The corner detection device according to claim 5, characterized in that, Also includes: The caching module is configured to cache the index information of each corner point in the corner point sequence, the sorted corner point sequence, and the updated corner point sequence.

7. The corner detection device according to claim 5, characterized in that, The corner suppression module is further configured to output the updated corner sequence to a memory outside the corner detection device.

Images