CN112991375A - Method and system for reshaping arbitrary-shaped image area into N rectangular areas - Google Patents
Method and system for reshaping arbitrary-shaped image area into N rectangular areas Download PDFInfo
- Publication number
- CN112991375A CN112991375A CN202110186850.5A CN202110186850A CN112991375A CN 112991375 A CN112991375 A CN 112991375A CN 202110186850 A CN202110186850 A CN 202110186850A CN 112991375 A CN112991375 A CN 112991375A
- Authority
- CN
- China
- Prior art keywords
- rectangle
- area
- matrix
- coordinate
- max
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000004364 calculation method Methods 0.000 claims abstract description 13
- 238000010008 shearing Methods 0.000 claims abstract description 8
- 238000005215 recombination Methods 0.000 claims abstract description 5
- 230000006798 recombination Effects 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 106
- 238000012545 processing Methods 0.000 claims description 27
- 230000008521 reorganization Effects 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 6
- 238000006116 polymerization reaction Methods 0.000 claims description 5
- 230000001131 transforming effect Effects 0.000 claims description 5
- 230000004931 aggregating effect Effects 0.000 claims description 2
- 239000000126 substance Substances 0.000 claims description 2
- 238000011161 development Methods 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 10
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000007429 general method Methods 0.000 description 1
- 238000007634 remodeling Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The application provides a method and a system thereof for reshaping an arbitrarily-shaped image area into N rectangular areas, wherein the method comprises the following steps: confirming a region of interest of an original image; marking the pixel of the region of interest as 1, and marking the pixels of other regions as 0; shearing the minimum external rectangles of the region of interest along the horizontal direction, and searching the rectangle with the largest area, namely the rectangle with the largest area according to a preset rule in the minimum external rectangles; taking the rectangle with the largest area as the center, and gradually and inwards converging other image pixels marked as 1 outside the rectangle with the largest area and inside the rectangle with the smallest external connection into the rectangle in a quicksand filling mode; if all or part of the data of the four outermost sides of the polymerized rectangle is not filled with 1, performing data recombination on the four outermost sides, and discarding the sides which are not filled after the data recombination; and cutting the processed rectangle into N rectangular areas. The method and the device facilitate the calculation of various algorithms for any region of interest, and accelerate the development of intelligent application.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a method and a system for reshaping an image area with any shape into N rectangular areas.
Background
In the field of computer vision, many algorithms are often required to be calculated for a picture, but the input of many algorithms is a matrix (a rectangular region in the picture), when many algorithms are calculated for a non-rectangular region of interest in the picture, the region cannot be directly used as the input of the algorithm, the edge can be filled into a rectangle for calculation, but the filled part and the region of interest have obvious changes, and the change, for example, as the input part of the algorithm, may seriously affect the output of the algorithm, so that the more general method is to select 1 to N regions in the region of interest to replace the target region for algorithm calculation. But at present, no more general algorithm is realized.
Disclosure of Invention
The invention aims to provide a method and a system for remolding an image area with any shape into N rectangular areas, which are convenient for carrying out multiple algorithm calculation on any interested area of an image so as to accelerate the development of intelligent application.
In order to achieve the purpose, the invention adopts the following technical scheme:
a first aspect of the present application provides a method for reshaping an arbitrarily-shaped image area into N rectangular areas, comprising:
acquiring an original image to be processed, and confirming an interested area of the original image, wherein the interested area is in any shape;
carrying out binarization processing on an original image, wherein the pixel of an interested area is marked as 1, and the pixels of other areas are marked as 0;
shearing the region of interest to obtain a minimum circumscribed rectangle of the region of interest along the horizontal direction;
searching a rectangle with the largest area in the cut minimum external rectangles according to a preset rule, wherein the rectangle is the rectangle with the largest area;
taking the rectangle with the largest area as the center, and gradually and inwards converging other image pixels marked as 1 outside the rectangle with the largest area and inside the rectangle with the smallest external connection into the rectangle in a quicksand filling mode;
when all or part of the data of the four outermost sides of the rectangle after polymerization is not filled with 1, performing data recombination on the four outermost sides, and filling the sides as much as possible; if the edge which is not filled with 1 still exists after the data is recombined, discarding the edge;
and cutting the processed rectangle into N rectangular areas, wherein N is a positive integer greater than 1.
Preferably, the cropping out the minimum bounding rectangle of the region of interest along the horizontal direction includes the following steps:
and (3) sequentially searching coordinates of 1 appearing for the first time from the left side to the inside of four edges at the outermost side of the image after binarization processing from the outside to the inside in a clockwise direction from the four edges, and sequentially recording the coordinates as Xmin、Ymin、Xmax、Ymax；
The coordinate of the upper left corner of the region of the smallest circumscribed rectangle is noted as (X)min，Ymin) The coordinate of the lower right corner is (X)max，Ymax)。
Preferably, in the cut minimum bounding rectangles, a rectangle with the largest area, that is, a rectangle with the largest area, is found according to a preset rule, and the method includes the following steps:
traversing the minimum circumscribed rectangle line by line from top to bottom;
each current row and the block of which the traversed row is continuously 1 in the y-axis direction form columns with different heights, and the height value is recorded;
searching each point from left to right as a maximum rectangle of the coordinate of the lower right corner which can be formed in the columnar graphs with different heights in the previous step;
calculating the area of the maximum rectangle, judging whether the area is the maximum or not, and updating the coordinate of the maximum rectangle if the area is the maximum;
and repeating the steps until all the rows are traversed to obtain the maximum area rectangular coordinate.
In a preferred embodiment, the corresponding matrix of the minimum bounding rectangle is a minimum bounding matrix A, with the coordinate of A being [ X [ ]min:Xmax,Ymin:Ymax]Wherein (X)min，Ymin) And (X)max，Ymax) The method comprises the following steps that the coordinates of the upper left corner and the lower right corner of a region of a minimum circumscribed rectangle are respectively, and the rectangle with the largest area is searched according to a preset rule, wherein the method comprises a matrix transformation step and an area calculation step; wherein the content of the first and second substances,
the matrix transforming step includes:
the minimum bounding matrix A is transformed into a height matrix B, each value in the height matrix B representing the maximum rectangular height that the position in the minimum bounding matrix A can make up in the y-axis direction, i.e. the maximum rectangular height
J represents the jth row of the matrix, i represents the ith column of the matrix, and i and j are both natural numbers more than or equal to 0; a. thejiRepresenting the pixel of the jth row and the ith column in the minimum circumscribed matrix A; b isjiThe pixel of the jth row and ith column in the height matrix B is represented;
the area calculating step includes:
according to each pixel B in the height matrix BjiSequentially calculating according to the following formula to obtain an area matrix S corresponding to the height matrix B, namely, taking each value in the area matrix S as a corresponding minimum external matrix A, taking pixel points at the same position as the coordinates of the lower right corner of a rectangle, summing the pixels of the largest rectangle searched upwards leftwards, and taking the pixel B as the pixel BjiThe calculation formula of (2) is as follows:
wherein S isjiThe pixel of the jth row and ith column in the area matrix S is represented; n, n1,n2,……nnAre all natural numbers more than or equal to 0;
finding the maximum value, i.e. the maximum area coordinate S, in the calculated area matrix Sji maxIf there are a plurality of values, the coordinate closest to the center of the area matrix S is taken, if there are still a plurality of valuesThen randomly taking one coordinate as the maximum area coordinate Sji max(ii) a In the minimum circumscribed matrix A, corresponding to the maximum area coordinate Sji maxThe pixel point of the position is the coordinate of the lower right corner of the rectangle, and the largest rectangle searched upwards leftwards is the rectangle with the largest area.
In a more preferred embodiment, in the process of calculating the area matrix S, the coordinates of the upper left corner of the maximum area rectangle are recorded through the sparse matrix K; in calculating the maximum area coordinate Sji maxWhen the maximum area coordinate S is used in the process of (2)ji maxThere are a plurality of values, one of which is randomly taken as the maximum area coordinate Sji maxThe calculation formula of (2) is as follows:
Sji max＝random(min(abs((Sj-Kj)/2-H/2))+abs((Si-Ki)/2-W/2))))
where H is the height of the matrix K, W is the width of the matrix K, KjMaximum area coordinate S recorded for matrix Kji maxLine coordinate, K, of the upper left corner of the corresponding maximum area rectangleiMaximum area coordinate S recorded for matrix Kji maxColumn coordinates, S, of the upper left corner of the corresponding maximum area rectanglejIs a maximum area coordinate Sji maxLine coordinate of (1), SiIs a maximum area coordinate Sji maxColumn coordinates of (a).
Preferably, the step of converging the image pixels marked as 1 outside the rectangle with the largest area and inside the rectangle with the smallest outline into a rectangle gradually inwards by a quicksand filling manner with the rectangle with the largest area as a center includes: starting from the left side of four edges at the outermost side of the maximum area rectangle, searching and filling the four edges outwards layer by layer in a clockwise direction; when filling the n-th side of the outer side of the maximum area rectangle, n is a positive integer larger than or equal to 1, the pixel with 1 in the n-th side is unchanged, the pixel with 0 is filled by the pixel with 1 in the nearest pixel of the (n +1) th side which is the adjacent outer side of the n-th side, and the nearest pixel is the pixel with the (n +1) th side and is in the range of closing and opening left and right by taking the maximum area rectangle as the center and forming an included angle of 45 degrees left and right.
In a preferred embodiment, the converging step by step inward into a rectangle by using a quicksand filling manner includes: the first column of the outer side of the maximum area rectangle is S from the left side of the four outermost sideskK is a natural number of 0 or more, then SkThe position of 1 in the column remains unchanged, SkThe position of 0 in the column is from the (S) thk-1) columns, and pixels having a lookup value of 1 among pixels within a left-closed and right-open range of an angle of 45 ° to the left and right thereof centered on the maximum area rectangle are filled in the S-th columnkColumn 0; and repeating the steps, and searching and filling all the data outwards layer by layer on the four sides of the outermost side of the rectangle with the largest area in a clockwise direction.
Preferably, after aggregating other pixels outside the rectangle with the largest area and inside the rectangle with the smallest external connection gradually inward into a rectangle by using a quicksand filling manner, performing data reorganization on four outermost sides to fill the most sides as much as possible, and discarding the side which is not filled with 1 if the side still exists after the data reorganization, including the following steps:
sequentially converging the data of four sides on the outermost side of the polymerized rectangle to the upper left corner point of the rectangle in the anticlockwise direction;
and sequentially checking the data of the four sides in the clockwise direction from the upper side of the four sides at the outermost side of the converged rectangle, reserving the sides filled with the data, and discarding the sides not filled with the data.
Preferably, the cutting the processed rectangle into N rectangular areas includes the following steps:
calculating the length of the short side, wherein the short side is the shorter side of the processed rectangle, and the longer side is the long side;
calculating the parts of the short side and the long side which need to be cut respectively as X and Y, wherein the rectangular area cut according to X, Y is the most approximate to a square;
calculating cutting coordinates according to X, Y, and performing rectangular cutting;
if the processed rectangle is a square, the short side and the long side are the sides of the square, and X is Y.
More preferably, the cutting the processed rectangle into N rectangular areas includes the following steps:
the length of the short side of the processed rectangle is h, the length of the long side is w, w > is h, and the number X of the parts needing to be cut on the short side and the number Y of the parts needing to be cut on the long side, X, Y are all positive integers which are more than or equal to 1 according to the following formula:
x Y + Z N, and the X, Y, Z values simultaneously satisfy the following constraints,
(1) z takes the value of 0 or 1;
(2)Y>＝X；
(3) (Y-X) is at a minimum in absolute value;
calculating the position coordinate of the short edge cutting, wherein the coordinate of the x part of the short edge cutting is as follows: x (h/X) and rounded (e.g., round rounding), X ═ 1, … …, X-1;
and (3) calculating the position coordinates of the long edge cutting, wherein when Z is 0, the coordinates of the y-th part of the long edge cutting are as follows: y (w/Y) and rounded (e.g., round), Y1, … …, Y-1; when Z is 1, the coordinates of the y' th part of the long edge cut are: y '(w/(Y +1)) and rounded (e.g., round), Y' ═ 1, … …, Y, i.e., when Z is 1, the number of cuts on the long side is actually (Y + 1).
In a second aspect, the present application provides a system for reshaping an arbitrarily-shaped image area into N rectangular areas, comprising:
the system comprises a region-of-interest confirming module, a region-of-interest confirming module and a processing module, wherein the region-of-interest confirming module is used for acquiring an original image to be processed and acquiring a region of interest of the original image, and the region of interest is in any shape;
the original image binarization processing module is used for carrying out binarization processing on the original image, wherein the pixels of the interested area are marked as 1, and the pixels of other areas are marked as 0;
the minimum circumscribed rectangle searching module is used for shearing the region of interest to obtain a minimum circumscribed rectangle of the region of interest along the horizontal direction;
the maximum area rectangle searching module is used for searching a rectangle with the maximum area in the cut minimum external rectangles according to a preset rule, namely the rectangle with the maximum area;
the quicksand type filling module is used for gradually inwards converging other image pixels marked as 1 outside the rectangle with the largest area and inside the rectangle with the smallest external connection into a rectangle by taking the rectangle with the largest area as the center;
the data reorganization module is used for reorganizing the data of the four outermost sides of the aggregated rectangle when all or part of the data of the four outermost sides of the aggregated rectangle is not filled with 1, and filling the sides as much as possible; if the edge which is not filled with 1 still exists after the data is recombined, discarding the edge;
and the rectangle cutting module is used for cutting the processed rectangle into N rectangular areas, wherein N is a positive integer greater than 1.
Compared with the prior art, the technical scheme of the invention has the following beneficial effects:
the method for remolding the image area in the any shape into the N rectangular areas can remold the interested image area in the picture into the N rectangular areas as algorithm input of multiple image processing, facilitates multiple algorithm calculation of the any interested area, improves accuracy of algorithm output of the image processing, effectively avoids distortion of the image processing, and accelerates development of intelligent application.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the application and, together with the description, serve to explain the application and are not intended to limit the application. In the drawings:
FIG. 1 is a flow chart of a method of reshaping an arbitrarily-shaped image region into N rectangular regions in accordance with the present invention;
FIG. 2 is an exemplary diagram of a method provided by an embodiment of the present invention for cropping an original image to form a minimum bounding rectangle;
FIG. 3 is a partial transformation example diagram of a least bounding rectangle with pixels transformed into a height matrix in a method provided by an embodiment of the invention;
FIG. 4 is a diagram illustrating an example of a processing result of transforming the minimum bounding matrix A to the height matrix B in the method according to the embodiment of the present invention;
FIG. 5 is a partially transformed example diagram of a height matrix B transformed to an area matrix S in a method provided by an embodiment of the invention;
fig. 6 is an exemplary diagram of a processing result of obtaining an area matrix S by calculating a height matrix B in the method according to the embodiment of the present invention;
FIG. 7 is a logic diagram illustrating an overall process of transforming from the minimum bounding matrix A to the area matrix S in the method provided by the embodiment of the invention;
FIG. 8 is a search direction and filling example of quicksand filling in the method provided by the embodiment of the present invention;
FIG. 9 is a diagram illustrating an example of data reorganization of the outermost four sides of a rectangle after aggregation in a method provided by an embodiment of the present invention;
fig. 10 is an example of rectangular cutting when Z is 0 in the method provided by the embodiment of the present invention;
fig. 11 is an example of rectangular cutting when Z is 1 in the method provided by the embodiment of the present invention.
Fig. 12 is a schematic structural diagram of a system for reshaping an arbitrarily-shaped image area into N rectangular areas according to an embodiment of the present invention.
Detailed Description
The purpose, technical scheme and effect of the invention are clearer and clearer, and the invention is further described in detail by referring to the attached drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Fig. 1 is a flowchart of a method for reshaping an arbitrarily-shaped image region into N rectangular regions according to the present invention. As shown in fig. 1, a method for reshaping an arbitrarily-shaped image area into N rectangular areas mainly includes the following steps:
step S01: acquiring an original image to be processed, and confirming an interested area of the original image to be processed, wherein the interested area is in any shape;
step S02: carrying out binarization processing on an original image to be processed, wherein the pixel of an interested area is marked as 1, and the pixels of other areas are marked as 0;
step S03: shearing the region of interest to obtain a minimum circumscribed rectangle of the region of interest along the horizontal direction;
step S04: searching a rectangle with the largest area, namely a rectangle with the largest area, in the cut minimum external rectangles according to a preset rule;
step S05: taking the rectangle with the largest area as the center, and gradually and inwards converging other image pixels marked as 1 outside the rectangle with the largest area and inside the rectangle with the smallest external connection into the rectangle in a quicksand filling mode;
step S06: when all or part of the data of the four outermost sides of the rectangle after polymerization is not filled with 1, performing data recombination on the four outermost sides, and filling the sides as much as possible; if the edge which is not filled with 1 still exists after the data is recombined, discarding the edge;
step S07: and cutting the processed rectangle into N rectangular areas, wherein N is a positive integer greater than 1.
Examples
Step S01: acquiring an original image to be processed, and confirming a region of interest of the original image to be processed, wherein the region of interest can be in any shape.
Step S02: and carrying out binarization processing on the original image, wherein the pixel of the region of interest is marked as 1, the pixels of other regions are marked as 0, and the coordinates of the region of interest are the processing results of other algorithms or the manually specified coordinates. The image pixel after the binarization processing is shown in fig. 2, in which a gray portion is a region of interest.
Step S03: shearing the interested region to obtain the minimum circumscribed rectangle of the interested region along the horizontal direction, and the steps are as follows:
and (3) sequentially searching coordinates of 1 appearing for the first time from the left side to the inside of four edges at the outermost side of the image after binarization processing from the outside to the inside in a clockwise direction from the four edges, and sequentially recording the coordinates as Xmin、Ymin、Xmax、Ymax；
Upper left and right of the area of the minimum bounding rectangleThe lower corner coordinates are respectively (X)min，Ymin) And (X)max，Ymax) That is, the corresponding matrix of the minimum bounding rectangle is the minimum bounding matrix a ═ Xmin:Xmax,Ymin:Ymax]. The processed minimum bounding rectangle is shown in FIG. 2 as the rectangle enclosed by the dashed line.
Step S04: in the cut minimum external matrix A, searching a rectangle with the largest area, namely a rectangle with the largest area according to a preset rule, and specifically comprising the following steps:
step S041: and (4) calculating a height matrix B.
And traversing the minimum circumscribed rectangle line by line from top to bottom, wherein each current line and a block of which the traversed line is continuously 1 in the y-axis direction form columns with different heights, and recording the height value. Referring to fig. 3, fig. 3 shows a partial transformation example of the height matrix. For example, in the minimum bounding rectangle, the pixel in the third row and the third column is 1, and the two pixels above it are both 1 (dotted frame portion), i.e., in a22At the point of time, A02、A12Both are 1, so the maximum rectangle height that the third row and the third column of pixels can form up is 3, and the corresponding third row and the third column of pixels in the height matrix is 3.
Specifically, the minimum bounding matrix a is converted into a height matrix B, each value in the height matrix B representing the maximum rectangular height that the position in the minimum bounding matrix a can be constructed to be directed upward in the y-axis direction, i.e., the maximum rectangular height
J represents the jth row of the matrix, i represents the ith column of the matrix, and i and j are both natural numbers more than or equal to 0; a. thejiRepresenting the pixel of the jth row and the ith column in the minimum circumscribed matrix A; b isjiWhich represents the pixel in the jth row and ith column of the height matrix B.
An exemplary diagram of the complete height matrix B after transformation of the minimum bounding matrix a is shown in fig. 4.
Step S042: and (4) calculating an area matrix S.
According to each pixel B in the height matrix BjiAnd sequentially calculating according to the following formula to obtain an area matrix S corresponding to the height matrix B:
wherein S isjiThe pixel of the jth row and ith column in the area matrix S is represented; n, n1,n2,……nnAre all natural numbers greater than or equal to 0.
Referring to fig. 5, fig. 5 shows a partial transformation example for transforming the height matrix B into the area matrix S. For example, in B22When pointing, first take [1, B22]Medium maximum positive integer, here 3, because B22The adjacent front (i.e. the left side thereof) does not have a value of 3 or more, i.e. n10, so S1(0+1) × 3 ═ 3 (that is, the rectangle formed by the pixels in the dashed box corresponding to the minimum circumscribed matrix a in fig. 5); then get (B)22-1), i.e. 2, because of B22The adjacent front (i.e. the left side thereof) does not have a value of 2 or more, i.e. n20, so S2(0+1) × (3-1) ═ 2; finally, get (B)22-2), i.e. 1, because of B22The adjacent front (i.e. the left side thereof) has 2 values of 1 or more in succession, i.e. n32, so S3The value is (2+1) × (3-2) ═ 3 (i.e., the rectangle formed by the pixels in the solid line frame in the minimum circumscribed matrix a in fig. 5). Namely, each value in the area matrix S is the sum of pixels of the largest rectangle found to the left and upward by taking the pixel point at the same position as the coordinate of the lower right corner of the rectangle in the corresponding minimum circumscribed matrix a.
An exemplary diagram of calculating the complete area matrix S from the height matrix B is shown in fig. 6.
Step S043: and recording coordinates of the upper left corner of the maximum area rectangle.
In the process of calculating the area matrix S in the previous step, the coordinates of the upper left corner of the rectangle with the largest area are recorded through the sparse matrix K.
Step S044: and calculating the maximum area coordinate.
Finding the maximum value, i.e. the maximum area coordinate S, in the calculated area matrix Sji maxIf there are a plurality of values, the coordinate closest to the center of the area matrix S is taken, and if there are still a plurality of values, one of the coordinates is randomly taken as the maximum area coordinate Sji max. In the minimum circumscribed matrix A, corresponding to the maximum area coordinate Sji maxThe pixel point of the position is the coordinate of the lower right corner of the rectangle, and the largest rectangle searched upwards leftwards is the rectangle with the largest area.
Specifically, the maximum area coordinate S is calculatedji maxWhen the maximum area coordinate S is used in the process of (2)ji maxThere are a plurality of values, one of which is randomly taken as the maximum area coordinate Sji maxThe calculation formula of (2) is as follows:
Sji max＝random(min(abs((Sj-Kj)/2-H/2))+abs((Si-Ki)/2-W/2))))
where H is the height of the matrix K, W is the width of the matrix K, KjMaximum area coordinate S recorded for matrix Kji maxLine coordinate, K, of the upper left corner of the corresponding maximum area rectangleiMaximum area coordinate S recorded for matrix Kji maxColumn coordinates, S, of the upper left corner of the corresponding maximum area rectanglejIs a maximum area coordinate Sji maxLine coordinate of (1), SiIs a maximum area coordinate Sji maxColumn coordinates of (a).
In this embodiment, the maximum area coordinate Sji maxIs S in the area matrix S in FIG. 745And 9, the maximum bounding matrix a can be mapped to a solid line box in the minimum bounding matrix a, and the solid line box is the searched maximum area rectangle.
Step S05: taking the rectangle with the largest area as the center, and gradually and inwards converging other image pixels marked as 1 outside the rectangle with the largest area and inside the rectangle with the smallest outside by adopting a quicksand filling mode into a rectangle, and specifically comprising the following steps of:
referring to fig. 8, in the minimum circumscribed matrix a, starting from the left side of the four outermost sides of the maximum area rectangle, the four sides are searched and filled outward layer by layer in a clockwise direction; when filling the n-th side of the outer side of the maximum area rectangle, n is a positive integer larger than or equal to 1, the pixel with 1 in the n-th side is unchanged, the pixel with 0 is filled by the pixel with 1 in the nearest pixel of the (n +1) th side which is the adjacent outer side of the n-th side, and the nearest pixel is the pixel with the (n +1) th side and is in the range of closing and opening left and right by taking the maximum area rectangle as the center and forming an included angle of 45 degrees left and right.
Specifically, the step-by-step inward polymerization into a rectangle by adopting a quicksand filling manner includes: the first column of the outer side of the maximum area rectangle is S from the left side of the four outermost sideskK is a natural number of 0 or more, then SkThe position of 1 in the column remains unchanged, SkThe position of 0 in the column is from the (S) thk-1) columns, and pixels having a lookup value of 1 among pixels within a left-closed and right-open range of an angle of 45 ° to the left and right thereof centered on the maximum area rectangle are filled in the S-th columnkColumn 0; and repeating the steps, and searching and filling all the data outwards layer by layer on the four sides of the outermost side of the rectangle with the largest area in a clockwise direction.
FIG. 8 shows the search direction and filling example of the quicksand filling, in which the solid line box represents the largest area rectangle and the curved dashed arrow represents the (S) thk-1) filling the S-th pixel with 1' S in the columnkThe position of the pixel in the column is 0 and the curved solid arrow indicates the seek direction looking outward, edge by edge, layer by layer, in a clockwise sequence.
Step S06: when all or part of the data of the outermost four sides of the rectangle after polymerization is not filled with 1, performing data reorganization on the outermost four sides to fill the most sides as far as possible, wherein the processing mode is as follows: referring to fig. 9, the data of the four outermost sides are sequentially converged toward the upper left corner point of the converged rectangle in the counterclockwise direction; and sequentially checking the data of the four sides in the clockwise direction from the upper side of the four sides at the outermost side of the converged rectangle, reserving the sides filled with the data, and discarding the sides not filled with the data.
Step S07: cutting the processed rectangle into N rectangular areas, wherein N is a positive integer greater than 1, and the method specifically comprises the following steps:
calculating the length of the short side, wherein the short side is the shorter side of the processed rectangle, and the longer side is the long side; calculating the parts of the short side and the long side which need to be cut respectively as X and Y, wherein the rectangular area cut according to X, Y is the most approximate to a square; calculating cutting coordinates according to X, Y, and performing rectangular cutting; if the processed rectangle is a square, the short side and the long side are the sides of the square, and X is Y.
Specifically, the length of the short side of the processed rectangle is h, the length of the long side is w, and w > is h, and the number of parts X of the short side to be cut and the number of parts Y of the long side to be cut, X, Y are all integers greater than or equal to 1 according to the following formula:
x Y + Z N, and the X, Y, Z values simultaneously satisfy the following constraints,
(1) z takes the value of 0 or 1;
(2)Y>＝X；
(3) (Y-X) is at a minimum in absolute value;
calculating the position coordinate of the short edge cutting, wherein the coordinate of the x part of the short edge cutting is as follows: x (h/X) and rounded (e.g., round rounding), X ═ 1, … …, X-1;
and (3) calculating the position coordinates of the long edge cutting, wherein when Z is 0, the coordinates of the y-th part of the long edge cutting are as follows: y (w/Y) and rounded (e.g., round), Y1, … …, Y-1; when Z is 1, the coordinates of the y' th part of the long edge cut are: y '(w/(Y +1)) and rounded (e.g., round), Y' ═ 1, … …, Y, i.e., when Z is 1, the number of cuts on the long side is actually (Y + 1).
Fig. 10 shows an example of the rectangular cutting when Z is 0, and referring to fig. 10, the length of the short side of the rectangle is h, and the length of the long side of the rectangle is w. The cutting mode is as follows: the rectangle is divided into 8 rectangular regions, i.e., N-8, where the short side is cut into 2 and the long side is cut into 4, i.e., X-2 and Y-4.
The coordinates of the first part of the short edge cut are: x is the number of1＝round(1*(h/2))；
The coordinates of the first part of the long edge cut are: y is1＝round(1*(w/4))；
The coordinates of the second cut of the long edge are: y is2＝round(2*(w/4))；
The coordinates of the third cut of the long edge are: y is3＝round(3*(w/4))。
Fig. 11 shows an example of rectangular cutting when Z is 1, and referring to fig. 11, the length of the short side of the rectangle is h, and the length of the long side of the rectangle is w. The cutting mode is as follows: the rectangle is divided into 7 rectangular regions, i.e., N-7, where the short side is cut into 2 parts and the long side is cut into 3 parts, i.e., X-2 and Y-3.
The coordinates of the first part of the short edge cut are: x is the number of1＝round(1*(h/2))；
The coordinates of the first part of the long edge cut are: y is1’＝round(1*(w/(3+1)))；
The coordinates of the second cut of the long edge are: y is2’＝round(2*(w/(3+1)))；
The coordinates of the third cut of the long edge are: y is3’＝round(3*(w/(3+1)))。
It can be seen that when Z is 1, the number of cut parts of the long side is actually 4, i.e., (Y + 1).
Referring to fig. 12, the present embodiment further provides a system 100 for reshaping an arbitrarily-shaped image area into N rectangular areas, comprising:
the region-of-interest confirming module 101 is configured to acquire an original image to be processed and acquire a region of interest of the original image, where the region of interest is in an arbitrary shape;
an original image binarization processing module 102, configured to perform binarization processing on an original image, where a pixel of an area of interest is marked as 1, and pixels of other areas are marked as 0;
the minimum circumscribed rectangle searching module 103 is used for shearing the region of interest to obtain a minimum circumscribed rectangle of the region of interest along the horizontal direction;
a maximum area rectangle searching module 104, configured to search, according to a preset rule, a rectangle with a maximum area from the cut minimum circumscribed rectangles, that is, a maximum area rectangle;
the quicksand type filling module 105 is used for gradually inwards converging other image pixels marked as 1 outside the rectangle with the largest area and inside the rectangle with the smallest external connection into a rectangle by taking the rectangle with the largest area as a center in a quicksand type filling mode;
the data reorganization module 106 is configured to, when all or part of the data of the outermost four sides of the aggregated rectangle is not filled with 1, reorganize the data of the outermost four sides, and fill the side with the largest amount as much as possible; if the edge which is not filled with 1 still exists after the data is recombined, discarding the edge;
and a rectangle cutting module 107, configured to cut the processed rectangle into N rectangular regions, where N is a positive integer greater than 1.
For the specific working process of the above modules, reference may be further made to the method for reshaping an image region with an arbitrary shape into N rectangular regions provided in the embodiments of the present invention, and this is not repeated here.
In summary, the present application provides a method and a system for remodeling an arbitrary-shaped image region into N rectangular regions, which can remodel an arbitrary-shaped region of interest in a picture into N rectangular regions as an input of algorithms for various image processing, thereby facilitating various algorithm calculations for arbitrary regions of interest, improving accuracy of algorithm output for image processing, and effectively avoiding distortion of image processing to accelerate development of intelligent application.
The embodiments of the present invention have been described in detail, but the embodiments are merely examples, and the present invention is not limited to the embodiments described above. Any equivalent modifications and substitutions to those skilled in the art are also within the scope of the present invention. Accordingly, equivalent changes and modifications made without departing from the spirit and scope of the present invention should be covered by the present invention.
Claims (10)
1. A method of reshaping an arbitrarily shaped image region into N rectangular regions, comprising:
acquiring an original image to be processed, and confirming an interested area of the original image, wherein the interested area is in any shape;
carrying out binarization processing on an original image, wherein the pixel of an interested area is marked as 1, and the pixels of other areas are marked as 0;
shearing the region of interest to obtain a minimum circumscribed rectangle of the region of interest along the horizontal direction;
searching a rectangle with the largest area in the cut minimum external rectangles according to a preset rule, wherein the rectangle is the rectangle with the largest area;
taking the rectangle with the largest area as the center, and gradually and inwards converging other image pixels marked as 1 outside the rectangle with the largest area and inside the rectangle with the smallest external connection into the rectangle in a quicksand filling mode;
when all or part of the data of the four outermost sides of the rectangle after polymerization is not filled with 1, performing data recombination on the four outermost sides, and filling the sides as much as possible; if the edge which is not filled with 1 still exists after the data is recombined, discarding the edge;
and cutting the processed rectangle into N rectangular areas, wherein N is a positive integer greater than 1.
2. A method for reshaping an arbitrarily-shaped image region into N rectangular regions as in claim 1, wherein said cropping a minimum bounding rectangle of the region of interest in the horizontal direction comprises the steps of:
and (3) sequentially searching coordinates of 1 appearing for the first time from the left side to the inside of four edges at the outermost side of the image after binarization processing from the outside to the inside in a clockwise direction from the four edges, and sequentially recording the coordinates as Xmin、Ymin、Xmax、Ymax；
The coordinate of the upper left corner of the region of the smallest circumscribed rectangle is noted as (X)min，Ymin) The coordinate of the lower right corner is (X)max，Ymax)。
3. A method for reshaping an arbitrarily-shaped image region into N rectangular regions as claimed in claim 1, wherein the step of finding a rectangle with a largest area, that is, a rectangle with a largest area, from the cut-out minimum bounding rectangles according to a preset rule, comprises the steps of:
traversing the minimum circumscribed rectangle line by line from top to bottom;
each current row and the block of which the traversed row is continuously 1 in the y-axis direction form columns with different heights, and the height value is recorded;
searching each point from left to right as a maximum rectangle of the coordinate of the lower right corner which can be formed in the columnar graphs with different heights in the previous step;
calculating the area of the maximum rectangle, judging whether the area is the maximum or not, and updating the coordinate of the maximum rectangle if the area is the maximum;
and repeating the steps until all the rows are traversed to obtain the maximum area rectangular coordinate.
4. A method for reshaping an arbitrarily-shaped image region into N rectangular regions as in claim 3, wherein the corresponding matrix of the smallest circumscribing rectangle is a smallest circumscribing matrix a having coordinates of [ X [ ]min:Xmax,Ymin:Ymax]Wherein (X)min，Ymin) And (X)max，Ymax) The method comprises the following steps that the coordinates of the upper left corner and the lower right corner of a region of a minimum circumscribed rectangle are respectively, and the rectangle with the largest area is searched according to a preset rule, wherein the method comprises a matrix transformation step and an area calculation step; wherein the content of the first and second substances,
the matrix transforming step includes:
the minimum bounding matrix A is transformed into a height matrix B, each value in the height matrix B representing the maximum rectangular height that the position in the minimum bounding matrix A can make up in the y-axis direction, i.e. the maximum rectangular height
Wherein j represents the jth row of the matrix, i represents the ith column of the matrix, and i and j are both largeA natural number equal to 0; a. thejiRepresenting the pixel of the jth row and the ith column in the minimum circumscribed matrix A; b isjiThe pixel of the jth row and ith column in the height matrix B is represented;
the area calculating step includes:
according to each pixel B in the height matrix BjiSequentially calculating according to the following formula to obtain an area matrix S corresponding to the height matrix B, namely, taking each value in the area matrix S as a corresponding minimum external matrix A, taking pixel points at the same position as the coordinates of the lower right corner of a rectangle, summing the pixels of the largest rectangle searched upwards leftwards, and taking the pixel B as the pixel BjiThe calculation formula of (2) is as follows:
wherein S isjiThe pixel of the jth row and ith column in the area matrix S is represented; n, n1,n2,……nnAre all natural numbers more than or equal to 0;
finding the maximum value, i.e. the maximum area coordinate S, in the calculated area matrix Sji maxIf there are a plurality of values, the coordinate closest to the center of the area matrix S is taken, and if there are still a plurality of values, one of the coordinates is randomly taken as the maximum area coordinate Sji max(ii) a In the minimum circumscribed matrix A, corresponding to the maximum area coordinate Sji maxThe pixel point of the position is the coordinate of the lower right corner of the rectangle, and the largest rectangle searched upwards leftwards is the rectangle with the largest area.
5. A method for reshaping an arbitrarily-shaped image region into N rectangular regions as in claim 4, wherein in the calculating of the area matrix S, the coordinates of the upper left corner of the largest-area rectangle are recorded by the sparse matrix K; in calculating the maximum area coordinate Sji maxWhen the maximum area coordinate S is used in the process of (2)ji maxThere are a plurality of values, one of which is randomly taken as the maximum area coordinate Sji maxThe calculation formula of (2) is as follows:
Sji max＝random(min(abs((Sj-Kj)/2-H/2))+abs((Si-Ki)/2-W/2))))
where H is the height of the matrix K, W is the width of the matrix K, KjMaximum area coordinate S recorded for matrix Kji maxLine coordinate, K, of the upper left corner of the corresponding maximum area rectangleiMaximum area coordinate S recorded for matrix Kji maxColumn coordinates, S, of the upper left corner of the corresponding maximum area rectanglejIs a maximum area coordinate Sji maxLine coordinate of (1), SiIs a maximum area coordinate Sji maxColumn coordinates of (a).
6. A method for reshaping an arbitrarily-shaped image area into N rectangular areas as claimed in claim 1, wherein said step of converging other image pixels labeled 1 outside and inside the largest-area rectangle into a rectangle by means of quicksand filling with the largest-area rectangle as a center comprises: starting from the left side of four edges at the outermost side of the maximum area rectangle, searching and filling the four edges outwards layer by layer in a clockwise direction; when filling the n-th side of the outer side of the maximum area rectangle, n is a positive integer larger than or equal to 1, the pixel with 1 in the n-th side is unchanged, the pixel with 0 is filled by the pixel with 1 in the nearest pixel of the (n +1) th side which is the adjacent outer side of the n-th side, and the nearest pixel is the pixel with the (n +1) th side and is in the range of closing and opening left and right by taking the maximum area rectangle as the center and forming an included angle of 45 degrees left and right.
7. A method for reshaping an arbitrarily shaped image area into N rectangular areas as claimed in claim 1, wherein after aggregating other pixels outside the largest-area rectangle and inside the smallest circumscribed rectangle into a rectangle step by step in a quicksand filling manner, performing data reorganization on the four outermost sides to fill the most sides as much as possible, and if there is still a side that is not filled with 1 after data reorganization, discarding the side, comprising the steps of:
sequentially converging the data of four sides on the outermost side of the polymerized rectangle to the upper left corner point of the rectangle in the anticlockwise direction;
and sequentially checking the data of the four sides in the clockwise direction from the upper side of the four sides at the outermost side of the converged rectangle, reserving the sides filled with the data, and discarding the sides not filled with the data.
8. A method of reshaping an arbitrarily shaped image area into N rectangular areas as in claim 1, wherein the cutting the processed rectangle into N rectangular areas comprises the steps of:
calculating the length of the short side, wherein the short side is the shorter side of the processed rectangle, and the longer side is the long side;
calculating the parts of the short side and the long side which need to be cut respectively as X and Y, wherein the rectangular area cut according to X, Y is the most approximate to a square;
calculating cutting coordinates according to X, Y, and performing rectangular cutting;
if the processed rectangle is a square, the short side and the long side are the sides of the square, and X is Y.
9. A method as claimed in claim 8, wherein the length of the short side of the processed rectangle is h, the length of the long side is w, w > -h, and the number of X parts to be cut on the short side and the number of Y parts to be cut on the long side, X, Y are all positive integers greater than or equal to 1 as follows:
X＊y + Z is N, and the X, Y, Z value simultaneously satisfies the following constraint,
(1) z takes the value of 0 or 1;
(2)Y>＝X；
(3) (Y-X) is at a minimum in absolute value;
calculating the position coordinate of the short edge cutting, wherein the coordinate of the x part of the short edge cutting is as follows: x is the number of＊(h/X) and rounded (e.g., round), X ═ 1, … …, X-1;
and (3) calculating the position coordinates of the long edge cutting, wherein when Z is 0, the coordinates of the y-th part of the long edge cutting are as follows: y is＊(w/Y) and roundingY-1, … …, Y-1; when Z is 1, the coordinates of the y' th part of the long edge cut are: y'＊(w/(Y +1)) and rounded, Y' is 1, … …, Y, i.e. when Z is 1, the number of cuts on the long side is actually (Y + 1).
10. A system for reshaping an arbitrarily shaped image region into N rectangular regions, comprising:
the system comprises a region-of-interest confirming module, a region-of-interest confirming module and a processing module, wherein the region-of-interest confirming module is used for acquiring an original image to be processed and acquiring a region of interest of the original image, and the region of interest is in any shape;
the original image binarization processing module is used for carrying out binarization processing on the original image, wherein the pixels of the interested area are marked as 1, and the pixels of other areas are marked as 0;
the minimum circumscribed rectangle searching module is used for shearing the region of interest to obtain a minimum circumscribed rectangle of the region of interest along the horizontal direction;
the maximum area rectangle searching module is used for searching a rectangle with the maximum area in the cut minimum external rectangles according to a preset rule, namely the rectangle with the maximum area;
the quicksand type filling module is used for gradually inwards converging other image pixels marked as 1 outside the rectangle with the largest area and inside the rectangle with the smallest external connection into a rectangle by taking the rectangle with the largest area as the center;
the data reorganization module is used for reorganizing the data of the four outermost sides of the aggregated rectangle when all or part of the data of the four outermost sides of the aggregated rectangle is not filled with 1, and filling the sides as much as possible; if the edge which is not filled with 1 still exists after the data is recombined, discarding the edge;
and the rectangle cutting module is used for cutting the processed rectangle into N rectangular areas, wherein N is a positive integer greater than 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110186850.5A CN112991375B (en) | 2021-02-08 | 2021-02-08 | Method and system for remolding image area with arbitrary shape into N rectangular areas |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110186850.5A CN112991375B (en) | 2021-02-08 | 2021-02-08 | Method and system for remolding image area with arbitrary shape into N rectangular areas |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112991375A true CN112991375A (en) | 2021-06-18 |
| CN112991375B CN112991375B (en) | 2024-01-23 |
Family
ID=76393366
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110186850.5A Active CN112991375B (en) | 2021-02-08 | 2021-02-08 | Method and system for remolding image area with arbitrary shape into N rectangular areas |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112991375B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116934774A (en) * | 2023-06-30 | 2023-10-24 | 安徽大学 | Quick and high-precision panoramic image clipping method |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080137987A1 (en) * | 2004-06-30 | 2008-06-12 | Keiji Kojima | Correcting image distortion caused by scanning |
| CN105957077A (en) * | 2015-04-29 | 2016-09-21 | 国网河南省电力公司电力科学研究院 | Detection method for foreign body in transmission lines based on visual saliency analysis |
| CN106898004A (en) * | 2017-01-04 | 2017-06-27 | 努比亚技术有限公司 | A kind of preprocess method for realizing interactive image segmentation, device and terminal |
| US20180092518A1 (en) * | 2015-06-24 | 2018-04-05 | Olympus Corporation | Image processing device, endoscope apparatus, information storage device, and image processing method |
| CN108205811A (en) * | 2016-12-16 | 2018-06-26 | 广东技术师范学院 | A kind of plant leaf blade minimum enclosed rectangle computational methods |
| CN108734131A (en) * | 2018-05-22 | 2018-11-02 | 杭州电子科技大学 | A kind of traffic sign symmetry detection methods in image |
| CN110782470A (en) * | 2019-11-04 | 2020-02-11 | 浙江工业大学 | Carpal bone region segmentation method based on shape information |
| CN111145261A (en) * | 2019-08-30 | 2020-05-12 | 广东星舆科技有限公司 | Method for identifying index point and storage medium |
| US20200402242A1 (en) * | 2017-12-26 | 2020-12-24 | Zhejiang Uniview Technologies Co., Ltd. | Image analysis method and apparatus, and electronic device and readable storage medium |
-
2021
- 2021-02-08 CN CN202110186850.5A patent/CN112991375B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080137987A1 (en) * | 2004-06-30 | 2008-06-12 | Keiji Kojima | Correcting image distortion caused by scanning |
| CN105957077A (en) * | 2015-04-29 | 2016-09-21 | 国网河南省电力公司电力科学研究院 | Detection method for foreign body in transmission lines based on visual saliency analysis |
| US20180092518A1 (en) * | 2015-06-24 | 2018-04-05 | Olympus Corporation | Image processing device, endoscope apparatus, information storage device, and image processing method |
| CN108205811A (en) * | 2016-12-16 | 2018-06-26 | 广东技术师范学院 | A kind of plant leaf blade minimum enclosed rectangle computational methods |
| CN106898004A (en) * | 2017-01-04 | 2017-06-27 | 努比亚技术有限公司 | A kind of preprocess method for realizing interactive image segmentation, device and terminal |
| US20200402242A1 (en) * | 2017-12-26 | 2020-12-24 | Zhejiang Uniview Technologies Co., Ltd. | Image analysis method and apparatus, and electronic device and readable storage medium |
| CN108734131A (en) * | 2018-05-22 | 2018-11-02 | 杭州电子科技大学 | A kind of traffic sign symmetry detection methods in image |
| CN111145261A (en) * | 2019-08-30 | 2020-05-12 | 广东星舆科技有限公司 | Method for identifying index point and storage medium |
| CN110782470A (en) * | 2019-11-04 | 2020-02-11 | 浙江工业大学 | Carpal bone region segmentation method based on shape information |
Non-Patent Citations (2)
| Title |
|---|
| KARTEEKA PAVAN KANADAM等: "Mammogram classification using sparse-ROI: A novel representation to arbitrary shaped masses", 《EXPERT SYSTEMS WITH APPLICATIONS》, pages 204 * |
| 王颖;王鹏飞;杨余旺;: "基于图像分割的目标尺寸特征测量", 计算机技术与发展, no. 02, pages 197 - 201 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116934774A (en) * | 2023-06-30 | 2023-10-24 | 安徽大学 | Quick and high-precision panoramic image clipping method |
| CN116934774B (en) * | 2023-06-30 | 2024-03-22 | 安徽大学 | Quick and high-precision panoramic image clipping method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112991375B (en) | 2024-01-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210201445A1 (en) | Image cropping method | |
| US8285077B2 (en) | Automatic correction of digital image distortion | |
| US8717632B2 (en) | Image processing method, image processing device and scanner | |
| CN107045634B (en) | Text positioning method based on maximum stable extremum region and stroke width | |
| CN111242954B (en) | Panorama segmentation method with bidirectional connection and shielding processing | |
| CN111353961B (en) | Document curved surface correction method and device | |
| CN108334879B (en) | Region extraction method, system and terminal equipment | |
| CN110060260B (en) | Image processing method and system | |
| CN114529925B (en) | Method for identifying table structure of whole line table | |
| JP2021135993A (en) | Text recognition method, text recognition apparatus, electronic device, and storage medium | |
| CN111598076B (en) | Method and device for detecting and processing date in label image | |
| CN111062331A (en) | Mosaic detection method and device for image, electronic equipment and storage medium | |
| CN112991375A (en) | Method and system for reshaping arbitrary-shaped image area into N rectangular areas | |
| CN111738040A (en) | Deceleration strip identification method and system | |
| WO2021147219A1 (en) | Image-based text recognition method and apparatus, electronic device, and storage medium | |
| JP2010002991A (en) | Image processor, image processing method, and computer program | |
| CN112507938A (en) | Geometric feature calculation method, geometric feature recognition method and geometric feature recognition device for text primitives | |
| JPWO2019041842A5 (en) | ||
| CN114511862B (en) | Form identification method and device and electronic equipment | |
| CN113408256B (en) | Table reconstruction method and device for table pictures and related equipment | |
| CN111178371B (en) | Target detection method, device and computer storage medium | |
| TWM623309U (en) | English font image recognition system | |
| CN113344948A (en) | Method for identifying small target by high-resolution picture | |
| CN114648751A (en) | Method, device, terminal and storage medium for processing video subtitles | |
| JP6116271B2 (en) | Feature amount calculation apparatus, method, and program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 200435 11th Floor, Building 27, Lane 99, Shouyang Road, Jing'an District, Shanghai Applicant after: Shanghai Tongban Information Service Co.,Ltd. Address before: 709, building a, Harui center, 555 Shouyang Road, Jing'an District, Shanghai 200072 Applicant before: Shanghai Tongban Information Service Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |