CN111667492A - Fruit image segmentation method based on radon transform - Google Patents
Fruit image segmentation method based on radon transform Download PDFInfo
- Publication number
- CN111667492A CN111667492A CN202010406347.1A CN202010406347A CN111667492A CN 111667492 A CN111667492 A CN 111667492A CN 202010406347 A CN202010406347 A CN 202010406347A CN 111667492 A CN111667492 A CN 111667492A
- Authority
- CN
- China
- Prior art keywords
- fruit
- image
- segmentation
- radon
- integral
- 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.)
- Pending
Links
- 235000013399 edible fruits Nutrition 0.000 title claims abstract description 156
- 229910052704 radon Inorganic materials 0.000 title claims abstract description 49
- SYUHGPGVQRZVTB-UHFFFAOYSA-N radon atom Chemical compound [Rn] SYUHGPGVQRZVTB-UHFFFAOYSA-N 0.000 title claims abstract description 49
- 238000000034 method Methods 0.000 title claims abstract description 20
- 238000003709 image segmentation Methods 0.000 title claims abstract description 17
- 230000011218 segmentation Effects 0.000 claims abstract description 31
- 230000009466 transformation Effects 0.000 claims abstract description 27
- 239000013598 vector Substances 0.000 claims description 36
- 239000006185 dispersion Substances 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 5
- 230000017105 transposition Effects 0.000 claims 1
- 238000012271 agricultural production Methods 0.000 abstract description 2
- 238000001514 detection method Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 6
- 241000220225 Malus Species 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 235000021016 apples Nutrition 0.000 description 2
- 238000005520 cutting process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 241000220324 Pyrus Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 235000021017 pears Nutrition 0.000 description 1
- 230000008569 process Effects 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
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/60—Rotation of whole images or parts thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/68—Food, e.g. fruit or vegetables
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)
Abstract
The invention relates to the technical field of fruit quality detection grading in computer image processing, computer vision and agricultural production, in particular to a fruit image segmentation method based on Radon transformation, which comprises the following steps: performing binarization processing on the fruit image, performing radon transformation, and calculating the optimal direction of fruit arrangement according to the radon transformation result R0 of the image; and performing corresponding rotation processing on the fruit image according to the calculated optimal direction, calculating the accurate segmentation position of each fruit in the image, and further performing fruit individual segmentation according to the segmentation position. The invention innovatively detects the optimal direction of continuous fruit arrangement, realizes image segmentation of the continuously arranged and conveyed fruits on the conveyor belt, and segments the continuously arranged fruits into individual independent fruit images.
Description
Technical Field
The invention relates to the technical field of fruit quality detection and classification in computer image processing, computer vision and agricultural production, in particular to a fruit image segmentation method based on Radon transformation.
Background
In order to improve the added value of fruit sales, the quality of the fruit needs to be graded correspondingly. With the continuous improvement of fruit grading technology, the grading technology is also developed from simple mechanical grading sorting to electronic grading sorting, and finally to grading sorting based on computer image processing technology and computer vision technology.
In fruit grading sorting equipment based on computer image processing technology and computer vision technology, a typical equipment structure comprises: one or more cameras, a fruit conveyor belt, a computer for analyzing the fruit grade, electromechanical means for sorting according to the fruit grading results. In such a sorting apparatus, in order to improve sorting efficiency in stages, fruits are arranged in a linear succession while being conveyed by a conveyor belt. Therefore, before performing a quality analysis on each fruit, it is necessary to segment the fruit arranged in succession: the method comprises the steps of using a camera to take pictures of the linear continuous arrangement of fruits on the conveyor belt, and dividing the continuous arrangement of fruit images into independent fruit individual images through a computer so as to grade and sort each fruit in the next quality.
However, the existing fruit grading and sorting equipment cannot effectively or accurately detect the arrangement direction of the continuous fruits, so that the positions of the individual continuous fruit division cannot be accurately calculated, and the accurate division of the individual continuous fruit division is finally influenced.
Disclosure of Invention
In order to solve the technical problems in the prior art, the invention provides a fruit image segmentation method based on Radon transformation, which innovatively detects the optimal direction of continuous fruit arrangement, realizes image segmentation of continuously arranged and conveyed fruits on a conveyor belt, and finally accurately segments the continuously arranged fruits into individual fruit images which are independent.
The invention is realized by adopting the following technical scheme: a fruit image segmentation method based on Radon transformation comprises the following steps:
s1, performing binaryzation processing on the fruit image, performing radon transformation, and calculating the optimal direction of fruit arrangement according to the radon transformation result R0 of the image;
and S2, performing corresponding rotation processing on the fruit image according to the obtained optimal direction, calculating the accurate segmentation position of each fruit in the image, and further performing segmentation according to the segmentation position.
Preferably, step S1 includes:
s11, converting the fruit image from the RGB color space to the HSV color space, and acquiring the value of a brightness V plane;
s12, performing binarization correction on the value of the brightness V plane;
s13, performing Radon transformation on the V plane after the binarization correction by taking the central point of the image as an original point, and recording an obtained Radon transformation result as R0;
and S14, calculating the optimal direction of the fruit arrangement according to the Radon transformation result R0.
Preferably, step S2 includes:
s21, setting the optimal fruit arrangement direction and the anticlockwise included Angle of the x axis of the image as Angle, and rotating the fruit image clockwise by the Angle;
s22, an integral vector corresponding to a straight line direction perpendicular to the optimal fruit arrangement direction is taken out from R0, that is, an integral vector corresponding to the (Angle +90) th mod 180 column is taken out from R0, and this integral vector is recorded as B.
And S23, calculating the segmentation position of the image according to the integral vector B, and segmenting the fruit individual according to the segmentation position.
Compared with the prior art, the invention has the following advantages and beneficial effects:
1. the invention innovatively detects the optimal direction of the continuously arranged fruits, thereby accurately calculating the cutting positions of the continuously arranged fruit individuals, solving the problem of accurately cutting the continuously arranged fruit individuals, and being applicable to various categories of fruits, including: oranges, apples, pears, etc.
2. The invention has good segmentation effect and certain tolerance to the noise of the image; the implementation mode is simple and easy to realize.
Drawings
FIG. 1 is a flow chart of the segmentation method of the present invention;
FIG. 2 is a view of a fruit in a near horizontal arrangement;
FIG. 3 is an image of fruit in an oblique arrangement;
FIG. 4 is a schematic view of the optimal linear direction of the fruit arrangement;
FIG. 5 is a value diagram of a V plane of an orange after binarization;
FIG. 6 is a schematic view of the best orientation of the fruit arrangement of FIG. 2;
FIG. 7 is a set of apple images arranged in series;
FIG. 8 is a value diagram of the V plane of an apple after binarization;
FIG. 9 is a schematic view of the preferred orientation of the arrangement of the fruits of FIG. 7;
FIG. 10 is a schematic representation of the results of the segmentation of the continuous fruit of FIG. 7;
fig. 11 is a schematic diagram of a two-dimensional matrix result obtained by radon transformation of an image.
Detailed Description
The present invention will be described in further detail with reference to examples and drawings, but the embodiments of the present invention are not limited thereto.
Examples
As shown in fig. 1, the fruit image segmentation method based on radon transform in this embodiment mainly includes: detecting the optimal direction of the fruit arrangement by radon transformation, performing corresponding rotation processing on the fruit image according to the detected optimal direction, taking out an integral vector corresponding to a direction straight line perpendicular to the optimal direction of the fruit arrangement from a radon transformation result R0, obtaining an optimal segmentation position and implementing segmentation.
First, detect the optimal direction of fruit arrangement
The method mainly comprises the steps of obtaining a fruit image, carrying out radon transformation after binarization processing is carried out on the image, and calculating the optimal direction of fruit arrangement according to the radon transformation result R0 of the image, wherein the method comprises the following specific steps:
(1) obtaining fruit images
When the conveyor belt carrying the fruits passes through the camera shooting range, the camera is started to shoot, the central area of the image is cut, and then the image is saved as an image file in a JPEG format. A typical image containing fruit in a linear succession is shown in figure 2.
(2) Making radon transform on image
And converting the fruit image from the RGB color space to the HSV color space to obtain the value of the brightness V plane. Further, performing binarization correction on the value of the brightness V plane: all values greater than or equal to 0.5 in the V plane are corrected to 1, and all values less than 0.5 are corrected to 0. And then performing Radon (Radon) transformation on the V plane after the binarization correction by taking the central point of the image area as an origin, taking 1 degree as a step length, anticlockwise from 0 degree to 179 degrees, and taking the distance of an integral straight line as a distance of 1 pixel. The obtained radon transform result is stored as a two-dimensional matrix, as shown in fig. 11: the two-dimensional matrix has 180 columns in total, the columns represent the anticlockwise included angle between the radon integral straight line and the x axis of the image, the values are 180 columns at intervals of 1 degree and from 0 degree to 179 degrees; the number of rows of the two-dimensional matrix is equal to the length of the image diagonal (rounded), for convenience of description, the length value of the diagonal is defined as L, the rows represent the distance from the radon integral straight line to the center point of the image, and the values are set at intervals of 1 pixel from-L/2 (rounded, negative sign indicates that the integral straight line is located on the left side of the origin) to L/2 (rounded, non-negative number indicates that the integral straight line is located on the right side of the origin). The result of the radon transform of the image is denoted as R0. Thus, each column of R0, i.e., column 0 to column 179, represents an integral vector formed by line integral values of L straight lines corresponding to angles (from 0 to 179 degrees), and the subscript of each column element, from bottom to top, represents the distance of the integral straight line of the corresponding angle from the center point of the image, and the element value represents the line integral value of the corresponding integral straight line.
(3) Calculating the optimal direction of fruit arrangement according to R0
In this embodiment, the "optimal direction of fruit arrangement" refers to the optimal straight line direction of coordinates of center points of fruits in the fitting image, and can be expressed as a counterclockwise angle between the fitting straight line and the x-axis of the image. For example, in the fruit image shown in fig. 3, the direction in which the fruits are arranged is shown as the direction indicated by the black straight line in fig. 4.
Since the individual fruit in the image has not been segmented at this time, the coordinates of the center point of each individual fruit in the image cannot be obtained, and thus the optimal direction cannot be calculated in accordance with the meaning of "optimal direction of fruit arrangement".
The method uses the result R0 of the radon transformation of the image to calculate the optimal direction of the fruit arrangement: in the optimal direction of fruit alignment, the integral vector of the radon transform has the lowest data dispersion. "data scatter" is a novel concept defined for convenience of description in the present invention. The data scatter is defined as follows:
let A ═ a1,a2,…,aN]T(T denotes a transpose) is a set of vectors composed of N elements, and assuming that there are M elements other than 0 in A, subscripts of the elements other than 0 are denoted by i1、i2、…、iMWherein M is less than or equal to N. The data dispersion degree Divergence of the vector A can be calculated by adopting the following calculation formula:
Center=(i1+i2+…+iM)/M
Divergence=[(i1-Center)2+(i2-Center)2+…+(iM-Center)2]/N
where Center is the subscript average of the M elements that are not 0. The data set with low data dispersion degree, the elements which are not 0 will be distributed on some adjacent positions in a concentrated way.
In order to calculate the optimal direction of the fruit arrangement, the data dispersion degree of each row of data of the image radon transform result R0, that is, each integral vector is calculated, and the angle corresponding to the integral vector with the smallest data dispersion degree is taken as the optimal direction of the fruit arrangement.
Taking the effective division of the oranges arranged in series as an example, the embodiment is further described as follows:
firstly, the fruit image shown in fig. 2 is converted from the RGB space to the HSV space, the value of the luminance V plane is obtained, and the V plane is binarized. The result of V-plane binarization is shown in fig. 5. In fig. 5, black dots represent 0 values, and white dots represent 1 values.
And performing Radon transformation on the binarized V plane by taking the central point of the image area as an original point, and calculating the optimal direction of continuous fruit arrangement in the fruit image according to the Radon transformation result. The preferred direction of fruit alignment for the continuous fruit shown in fig. 2 is shown in fig. 6.
And secondly, performing corresponding rotation processing on the fruit image according to the calculated optimal direction, calculating the optimal segmentation position of the fruit, and further performing segmentation according to the obtained segmentation position.
As shown in fig. 7 to 10, in the fruit image segmentation method based on radon transform according to the present embodiment, after the optimal direction of the fruit arrangement is calculated, the fruit image is rotated, and then the optimal segmentation position of each fruit is calculated and segmented. The embodiment mainly comprises the steps of rotating a fruit image, taking out an integral vector corresponding to a linear direction perpendicular to the optimal fruit arrangement direction from R0, calculating the image segmentation position and segmenting fruit individuals, and the specific steps comprise:
(1) rotating fruit images
And setting the optimal counterclockwise included Angle between the direction and the x axis of the image as Angle, and rotating the fruit image clockwise by the Angle.
(2) An integral vector corresponding to a linear direction perpendicular to the optimal fruit arrangement direction is taken out from R0
An integral vector corresponding to a straight line direction perpendicular to the optimal direction of the fruit is taken out from R0, namely an integral vector corresponding to the (Angle +90) th mod 180 column is taken out from R0, and the integral vector is recorded as B.
(3) Constructing the segmentation position of the image according to the integral vector B, and segmenting the fruit individual
Since the integral vector B is obtained by line-integrating L straight lines perpendicular to the optimal fruit alignment direction, the element values of the integral vector B have the following characteristics: in the existing area of fruit in the image, the line integral value is large; in the adjacent area of the fruit, the line integral value is small and even 0; in the existing area of each fruit, the change of the B element value is represented as a change process from small to large and then from large to small; in adjoining areas of the fruit, valleys of line integrals can occur. Therefore, the dividing position of the fruit individual can be determined only by finding the trough of the data change from the integral vector B.
Based on such observation, the elements whose element values at the beginning and end are continuously 0 are first deleted from the integral vector B: set at the beginning to delete K1Elements, with K deleted at the end2An element; the integral vector B is then scanned from bottom to top, recording the position of each trough relative to the lowest element of the vector B: is provided with Q wave troughs, and the positions of the wave troughs are P respectively1、P2、…、PQThen, the segmentation position of each individual fruit in the fruit image is: the 1 st fruit is divided into [ K ]1,P1]The 2 nd fruit is divided into [ P ]1,P2]By analogy, the last fruit division position [ P ]Q,L-K2]. And segmenting the rotated image according to the obtained fruit segmentation interval, and returning to segment to obtain each fruit individual image.
Taking the effective division of the apples arranged in series as an example, the embodiment is further described as follows:
first, the fruit image shown in fig. 7 is converted from RGB space to HSV space, the value of the luminance V-plane is obtained, and the V-plane is binarized, and the result of binarization is shown in fig. 8, where black dots represent 0 values and white dots represent 1 values.
And performing Radon transformation on the binarized V plane by taking the central point of the image area as an original point, and calculating the optimal direction of continuous fruit arrangement in the fruit image according to the result data of the Radon transformation. The preferred direction of fruit alignment for the continuous fruit shown in fig. 7 is shown in fig. 9. The optimal direction of fruit alignment is at an angle of 17 degrees counterclockwise from the x-axis of the image.
The fruit image is rotated by 17 degrees clockwise, and an integral vector corresponding to the (90+17) th mod 180 which is 107 degrees, namely 107 th row data in R0, is taken out from R0, and the division position of each fruit in the continuous fruit arrangement is calculated: a first fruit position [32,80], a second fruit position [80,122], a third fruit position [122,169], a fourth fruit position [169,217], and a fifth fruit position [217,274 ]. The result of the segmentation is shown in fig. 10.
This embodiment can effectively segment the fruits with noise in a continuous arrangement, for example, a fruit image formed by adding white gaussian noise with a mean of 0 and a variance of 0.005 to the image, and the above segmentation steps can be applied to complete the segmentation of each individual fruit.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.
Claims (6)
1. A fruit image segmentation method based on Radon transform is characterized by comprising the following steps:
s1, performing binaryzation processing on the fruit image, performing radon transformation, and calculating the optimal direction of fruit arrangement according to the radon transformation result R0 of the image;
and S2, performing corresponding rotation processing on the fruit image according to the obtained optimal fruit arrangement direction, calculating the accurate segmentation position of each fruit in the image, and further performing segmentation according to the segmentation position.
2. The radon transform-based fruit image segmentation method as set forth in claim 1, wherein the step S1 includes:
s11, converting the fruit image from the RGB color space to the HSV color space, and obtaining the value of a brightness V plane;
s12, performing binarization correction on the brightness V plane;
s13, performing Radon transformation on the V plane after the binarization correction by taking the central point of the image as an original point, and recording an obtained Radon transformation result as R0;
and S14, calculating the optimal direction of the fruit arrangement according to the Radon transformation result R0.
3. The radon transform-based fruit image segmentation method as set forth in claim 2, wherein the step S14 includes:
let A ═ a1,a2,…,aN]TIs a vector composed of N elements, and M elements which are not 0 are arranged in the vector A, and the subscripts of the elements which are not 0 are respectively i1、i2、…、iMWherein T represents transposition, M is less than or equal to N; calculating the data dispersity Divergence of the vector A:
Center=(i1+i2+…+iM)/M
Divergence=[(i1-Center)2+(i2-Center)2+…+(iM-Center)2]/N
wherein Center is the subscript average of the M elements other than 0;
storing the radon transform result R0 as a two-dimensional matrix, wherein the column represents the counterclockwise included angle between the radon integral straight line and the x axis of the image, and the row represents the distance between the radon integral straight line and the central point of the image; and calculating the data dispersion degree of each row of integral vectors of the radon transform result R0, and taking the angle corresponding to the integral vector with the minimum data dispersion degree as the optimal direction of fruit arrangement.
4. The radon transform-based fruit image segmentation method as set forth in claim 2, wherein the step S12 binarizes and corrects the luminance V plane into: all values greater than or equal to 0.5 in the V plane are corrected to 1, and all values less than 0.5 are corrected to 0.
5. The radon transform-based fruit image segmentation method as set forth in claim 1, wherein the step S2 includes:
s21, setting the optimal fruit arrangement direction and the anticlockwise included Angle of the x axis of the image as Angle, and rotating the fruit image clockwise by the Angle;
s22, taking an integral vector corresponding to a straight line direction perpendicular to the optimal fruit arrangement direction from R0, namely taking an integral vector corresponding to the (Angle +90) th mod 180 column from R0, and recording the integral vector as B;
and S23, constructing the segmentation position of the image according to the integral vector B, and segmenting the fruit individual according to the segmentation position.
6. The radon transform-based fruit image segmentation method as set forth in claim 5, wherein the step S23 includes:
elements with continuous 0 values at the beginning and end are first deleted from the integral vector B: set at the beginning to delete K1Elements, with K deleted at the end2An element; the integral vector B is then scanned from bottom to top, recording the position of each trough relative to the lowest element of the integral vector B: is provided with Q wave troughs, and the positions of the wave troughs are P respectively1、P2、…、PQThen, the segmentation position of each individual fruit in the fruit image is: the 1 st fruit is divided into [ K ]1,P1]The 2 nd fruit is divided into [ P ]1,P2]By analogy, the last fruit division position [ P ]Q,L-K2](ii) a And segmenting the rotated image according to the obtained fruit segmentation interval, and returning each fruit individual image obtained by segmentation.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010406347.1A CN111667492A (en) | 2020-05-14 | 2020-05-14 | Fruit image segmentation method based on radon transform |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010406347.1A CN111667492A (en) | 2020-05-14 | 2020-05-14 | Fruit image segmentation method based on radon transform |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111667492A true CN111667492A (en) | 2020-09-15 |
Family
ID=72383625
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010406347.1A Pending CN111667492A (en) | 2020-05-14 | 2020-05-14 | Fruit image segmentation method based on radon transform |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111667492A (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2011100575A2 (en) * | 2010-02-11 | 2011-08-18 | Emory University | Systems, methods and computer readable storage mediums storing instructions for applying multiscale bilateral filtering to magnetic resonance (mr) images |
CN105023013A (en) * | 2015-08-13 | 2015-11-04 | 西安电子科技大学 | Target detection method based on local standard deviation and Radon transformation |
CN109447067A (en) * | 2018-10-24 | 2019-03-08 | 北方民族大学 | A kind of bill angle detecting antidote and automatic ticket checking system |
-
2020
- 2020-05-14 CN CN202010406347.1A patent/CN111667492A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2011100575A2 (en) * | 2010-02-11 | 2011-08-18 | Emory University | Systems, methods and computer readable storage mediums storing instructions for applying multiscale bilateral filtering to magnetic resonance (mr) images |
CN105023013A (en) * | 2015-08-13 | 2015-11-04 | 西安电子科技大学 | Target detection method based on local standard deviation and Radon transformation |
CN109447067A (en) * | 2018-10-24 | 2019-03-08 | 北方民族大学 | A kind of bill angle detecting antidote and automatic ticket checking system |
Non-Patent Citations (2)
Title |
---|
吴学品: "《直觉与统计》", 吉林大学出版社, pages: 27 - 30 * |
祁平: "基于支持向量机的车牌识别***的研究", pages 63 - 32 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11429818B2 (en) | Method, system and device for multi-label object detection based on an object detection network | |
Qureshi et al. | A bibliography of pixel-based blind image forgery detection techniques | |
US5647027A (en) | Method of image enhancement using convolution kernels | |
US9275305B2 (en) | Learning device and method, recognition device and method, and program | |
CN111008961B (en) | Transmission line equipment defect detection method and system, equipment and medium thereof | |
CN107038416B (en) | Pedestrian detection method based on binary image improved HOG characteristics | |
Singh et al. | Adaptive binarization of severely degraded and non-uniformly illuminated documents | |
CN110335233B (en) | Highway guardrail plate defect detection system and method based on image processing technology | |
US20140301608A1 (en) | Chemical structure recognition tool | |
Sheshkus et al. | HoughNet: neural network architecture for vanishing points detection | |
US11348349B2 (en) | Training data increment method, electronic apparatus and computer-readable medium | |
CN110070545B (en) | Method for automatically extracting urban built-up area by urban texture feature density | |
CN115019160B (en) | Method for controlling image quality of warehouse goods in factory | |
CN112200019A (en) | Rapid building night scene lighting light fault detection method | |
CN110648330A (en) | Defect detection method for camera glass | |
CN111597875A (en) | Traffic sign identification method, device, equipment and storage medium | |
CN110569848A (en) | feature extraction method and system for power equipment nameplate | |
CN111914847B (en) | OCR (optical character recognition) method and system based on template matching | |
WO2022121021A1 (en) | Identity card number detection method and apparatus, and readable storage medium and terminal | |
Jetley et al. | Two-stage hybrid binarization around fringe map based text line segmentation for document images | |
Lech et al. | Prediction of the optical character recognition accuracy based on the combined assessment of image binarization results | |
US6154565A (en) | Automatic retrieval process for machine-printed or handwritten text on a background, in a multilevel digital image | |
CN111667492A (en) | Fruit image segmentation method based on radon transform | |
CN117011259A (en) | Fruit defect detection method, device, equipment and medium | |
Yang et al. | License plate detection based on sparse auto-encoder |
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 |