CN110619648A - Method for dividing image area based on RGB change trend - Google Patents
Method for dividing image area based on RGB change trend Download PDFInfo
- Publication number
- CN110619648A CN110619648A CN201910887174.7A CN201910887174A CN110619648A CN 110619648 A CN110619648 A CN 110619648A CN 201910887174 A CN201910887174 A CN 201910887174A CN 110619648 A CN110619648 A CN 110619648A
- Authority
- CN
- China
- Prior art keywords
- similarity
- rgb
- array
- pixel
- dimensional
- 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
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
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- 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
- G06T7/90—Determination of colour characteristics
-
- 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
- G06T2207/10024—Color image
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for dividing image areas based on RGB (red, green and blue) change trends, which mainly utilizes two rounds of similarity judgment to divide the image areas, utilizes the similarity of change directions of RGB three-primary-color channels and the physical position proximity of pixel points to judge whether the pixel points are one of certain area clusters or not, and belongs to a picture preprocessing method. The method for dividing the image area based on the RGB change trend can judge according to the change trend of the color signal, so that the edge division is more definite, and the accuracy of the image area division is effectively improved.
Description
Technical Field
The invention relates to the technical field of image object recognition preprocessing, in particular to a method for dividing image areas based on RGB (red, green and blue) change trends.
Background
Image segmentation is a basic computer vision technique, and is a key step from image processing to image analysis. Efficient and rational image segmentation can abstract very useful information for content-based image retrieval, object analysis, and the like, thereby enabling higher-level image understanding. Image segmentation is still a problem that has not been solved well to date and is still a hot issue of research.
Most of the existing image segmentation methods convert an image to be processed into a gray image, determine the similarity of image pixel points based on a threshold value by utilizing the direct difference of the gray values of the image, and judge the image boundary so as to divide the region of the image. Compared with classical algorithms, the algorithms include threshold-based segmentation (fixed threshold segmentation, adaptive threshold image segmentation and the like), edge-based segmentation (Canny edge detection and the like), region-based segmentation (a seed region growing method, a region splitting and merging method and the like), and the algorithms lack color information, so that regions with the same brightness but less obvious color change cannot be distinguished, the edge detection is inaccurate, and the divided regions are inaccurate. Even if a few algorithms for performing region division based on color threshold values exist, a single color difference threshold value is used as a judgment standard, the influence of local information of an image on human eye color difference perception is neglected by the selection of the threshold value, and the judgment of a boundary is difficult, so that the region division is inaccurate.
Disclosure of Invention
The present invention is directed to overcome the above-mentioned deficiencies in the background art, and provides a method for dividing an image area based on RGB variation trend, which can determine the RGB variation trend according to the variation trend of a color signal, so that edge division is more definite, and the accuracy of image area division is effectively improved.
In order to achieve the technical effects, the invention adopts the following technical scheme:
a method for dividing image areas based on RGB change trend comprises the following steps:
A. acquiring an input picture to obtain an image with the length of X pixels and the width of Y pixels; the input format and the pixel size are not specially limited, PNG, JPEG and the like can be adopted, and meanwhile, for convenience of calculation, the pixel size is preferably the same as the length and width of the pixel;
B. acquiring RGB three-channel signal values of pixel points and converting the RGB three-channel signal values into two-dimensional arrays, wherein the length of the arrays is X X Y, the specific array values are [ … [ R, G, B ], [ R, G, B ] … ], and each pixel group comprises three elements to represent the RGB signal values of the pixel points;
C. the method comprises the steps of obtaining the slope of a line segment formed by pixel point color signals, filling the group of signals into three points [0, R ], [1, G ], [2, B ] ] with one pixel signal as a unit, calculating the slopes of two line segments formed by the three points, specifically forming a line segment I passing the points [0, R ], [1, G ] ] and a line segment II passing the points [1, G ], [2, B ], calculating the slopes K1 of the line segment I and the slopes K2 of the line segment II, K1 ═ R-G, K2 ═ B-G, obtaining the slope array of each pixel point relative to a color channel (capable of reflecting the change trend of the pixel point relative to an RGB signal), and obtaining the final slope array [ GN-RN, BN-GN ], [ BN-GN, BN- … ], [ GN-GN ] of each pixel point relative to the change situation of the color signals, is marked as [ [ slopeA, slopeB ], [ slopeA, slopeB ] … ];
the slope of the straight line represents the change condition of the straight line, the direction vector of the slope represents the change direction of the straight line, and similarly, the slope direction of the line segment formed by the color signals represents the change direction of the color signals of the pixel points, and the change directions are similar, so that the color signals of the pixel points are similar, otherwise, the color signals are not similar;
D. performing a first round of aggregation, comparing the change trends of the RGB signals in the neighborhood of the pixel points, and dividing the slopes in the same direction in the slope array into the same type, namely judging the slopes to belong to the same region; the concrete embodiment is that the slopes at the same position are all positive or negative;
E. performing a second round of aggregation, calculating and comparing the similarity of RGB color signals between the regions defined in the first round, judging whether two regions with certain similarity are combined according to a preset similarity threshold, and combining the regions which can be combined, wherein the specific method for calculating the similarity can adopt any one similarity algorithm in the prior art, and the preset similarity threshold can be set as a fixed value, and can also be adaptively adjusted according to the overall threshold of the image;
F. and outputting the final combined region judgment result.
Further, the step E specifically includes:
E1. mapping the [ R, G, B ] signals of the single pixels of each region to a one-dimensional space to obtain corresponding one-dimensional values, and combining the one-dimensional values of the single pixels of each region to form a one-dimensional array of the region;
E2. respectively sequencing and sampling the one-dimensional arrays of every two adjacent regions, calculating the similarity between the two arrays of every two adjacent regions and obtaining the maximum similarity between all the regions;
E3. multiplying the maximum similarity by the distance factor to obtain a final similarity; the distance factor a is (m-n)/m, wherein m is the maximum distance of the image, and a is the minimum distance between the two areas;
E4. and comparing whether the final similarity exceeds a preset similarity threshold, if so, judging that the two areas can be merged, otherwise, not merging.
Further, when the one-dimensional arrays are sorted and sampled in step E2, the numerical values in each array are sorted in descending order, and then the numbers of the numbers included in the two arrays are compared, and the first n numbers are reserved for the array with the larger number of numbers, where n is the number of the numbers included in the array with the smaller number of numbers.
Further, in step E2, the similarity between the number groups is calculated by specifically using a cosine similarity calculation method, and the calculation formula is as follows:
wherein Xi represents a one-dimensional value of each pixel point of one of the two compared arrays, and Yi represents a one-dimensional value of each pixel point of the other array.
Compared with the prior art, the invention has the following beneficial effects:
according to the method for dividing the image areas based on the RGB change trend, the RGB color signals of the image are firstly utilized, area grouping is carried out according to the change trend of the color signals, then the area similarity is utilized for combination, the purpose of final area division is achieved, the image areas can be divided more accurately based on the given image, useful information is abstracted for content-based image retrieval, object analysis and the like, high-level image understanding is facilitated, and the method has the advantages of simplicity, effectiveness and accurate dividing result.
Drawings
Fig. 1 is a schematic flow chart of the method for dividing image areas based on RGB variation tendency according to the present invention.
Fig. 2 is a schematic diagram of artwork input in one embodiment of the present invention.
FIG. 3 is a schematic view of a polyline plotted from the calculated slope aggregation in one embodiment of the present invention.
Detailed Description
The invention will be further elucidated and described with reference to the embodiments of the invention described hereinafter.
Example (b):
the first embodiment is as follows:
as shown in fig. 1, a method for dividing an image region based on RGB variation trend mainly uses two rounds of similarity determination to divide the image region, and determines whether a pixel point is one of a certain region cluster by using the similarity of the variation directions of RGB three-primary color channels and the proximity of the physical position of the pixel point, belonging to a picture preprocessing method.
Specifically, in this embodiment, the method for dividing the image area based on the RGB variation trend specifically includes the following steps:
step s01, accepting input of a picture, where the picture input in this embodiment is the original as shown in fig. 2, specifically, the image shown in fig. 2 is an image with the original being decolored, and in this embodiment, image information with length X equal to 91 pixels and width Y equal to 27 pixels is obtained from the image.
And S02, reading and storing RGB color channel signal values of the pixel points in the example diagram of FIG. 2 to obtain a [ … [ R, G, B ], [ R, G, B ] … ] color channel two-dimensional array.
Specifically, the length of the array obtained in this embodiment is X × Y ═ 91 × 27 ═ 2457, and specific examples of the array are as follows (denoted by data D1):
[…[83,108,144],[11,51,152],[24,64,203],[19,62,195],[16,69,187],[16,63,192],[245,255,252]…]
each pixel group contains three elements representing the RGB signal value of the pixel, for example [83,108,144] represents the RGB signal value of a certain pixel.
Step s03, calculating color signal slopes, where each group of signals includes 3 points, and taking a pixel RGB signal [83,108,144] in example data D1 as an example, the 3 points that can be obtained by filling are [ [0,83], [1,108], [2,144] ], and then two by two are sequentially formed into a straight line, and 2 straight lines are total, then the straight line passes through the point [0,83] and the point [1,108], and the straight line passes through the point [1,108] and the point [3,144], and then calculating slopes of two line segments, where the slopes of the two line segments obtained according to a slope formula K ═ Y2-Y2)/(X2-X1 are respectively K1 (108-83)/(1-0) ═ 25, K2 ═ 144-.
Then, the RGB signals of each pixel in the sample data D1 are processed and calculated in the above processing manner, so as to obtain a final slope array (denoted as data D2) corresponding to the sample data D1 and related to the color signal change condition:
[…[25,36],[40,101],[40,139],[43,133],[53,118],[47,129],[10,-3]…]
the corresponding slope distribution graph can be obtained by plotting points according to the obtained slope, and as shown in fig. 3, the slope distribution graph is made according to the calculated slope plotting points in this embodiment, and it can be visually seen that the regions are composed of the same slope direction and the different slope directions.
Step s04, performing a first round of region division, and comparing the change trends of the RGB signals in the neighborhood of the pixel point, in this example, comparing the change trends of the pixels in the neighborhood of the pixel point 8 (where the size of the neighborhood is not specially limited), to determine whether the neighborhood pixel points are in the same group, and when the change trends of the neighborhood pixel points are in the same direction, the neighborhood pixel points are considered to belong to the same region, for example, in the data D2 in this embodiment, [25,36], [40,101], [40,139], [43,133], [53,118], [47,129] are regarded as the same region in the same category, and [ 10-3 ] are regarded as a separate region in the other category.
And S05, carrying out region combination of the second round, comparing the region similarity, and carrying out combination.
Specifically, the data is first mapped to a one-dimensional space, in this example, [83,108,144] in the example data D1 is mapped to the one-dimensional space, and since the RGB color is 8 bits and takes a value of 0 to 255, the result of mapping the array to the one-dimensional space is 83 × 256+108 × 256+144 — 5424827.
Then, one-dimensional values of the pixel points in the region are calculated in sequence to obtain a one-dimensional array related to region color information, and then, the one-dimensional array is a mathematical problem, namely, the similarity of unordered indefinite-length arrays is compared, for example, two groups of existing one-dimensional arrays are provided, wherein array 1 is [1,1,2,1,1,1,0,0,0]The array 2 is [1,1,1,0,1,1,1,1,2,5,6,7 ]](two relatively simple arrays are illustrated in this embodiment to explain the subsequent specific calculation process), after sampling (i.e., sorting is to sort the numbers in each array in the order of the numbers from small to large) to obtain the maximum similarity array, i.e., the array to be compared after sampling is [0,0,0,1,1,1, 2 ], where the number of the arrays is small in number and the number of the arrays is large in number is intercepted to obtain the maximum similarity array]And [0,1,1,1,1,1,1,1,2]The array is substituted into the cosine similarity formula,(wherein i is 9, X1＝0，X2＝0，X2＝0，X4＝1，…X9＝2；Y1＝0，Y2＝1，Y2＝1，Y4＝1，…Y92), the finally obtained similarity is 0.9, and the obtained similarity is multiplied by a distance factor, in this embodiment, the distance of the region is 2 pixels, and the maximum distance of the image is 100 pixels, the distance factor obtained by the ratio is (100-2)/100, that is, 98%, and therefore, the finally obtained similarity is 0.9 × 0.98 — 0.88.
And finally, judging whether the two areas with certain similarity are combined or not according to a preset similarity threshold, comparing whether the final similarity exceeds the preset similarity threshold, if so, judging that the two areas can be combined, otherwise, not combining. The preset similarity threshold set in this embodiment is a fixed threshold of 0.75, and since 0.88 is greater than 0.75, the two areas where the array 1 and the array 2 are located can be merged.
And S06, outputting a final area judgment result through two rounds of area division.
It will be understood that the above embodiments are merely exemplary embodiments taken to illustrate the principles of the present invention, which is not limited thereto. It will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the spirit and substance of the invention, and these modifications and improvements are also considered to be within the scope of the invention.
Claims (4)
1. A method for dividing image areas based on RGB change trend is characterized by comprising the following steps:
A. acquiring an input picture to obtain an image with the length of X pixels and the width of Y pixels;
B. acquiring RGB three-channel signal values of pixel points and converting the RGB three-channel signal values into two-dimensional arrays, wherein the length of the arrays is X X Y, the specific array values are [ … [ R, G, B ], [ R, G, B ] … ], and each pixel group comprises three elements to represent the RGB signal values of the pixel points;
C. obtaining the slope of the line segment formed by the pixel point color signals, specifically taking a pixel signal as a unit [ R, G, B ], filling the group of signals into three points [ [0, R ], [1, G ], [2, B ] ], calculating the slopes of two line segments formed by three points, specifically forming a first line segment which respectively crosses the point [ [0, R ], [1, G ] ] and a second line segment which respectively crosses the point [ [1, G ], [2, B ] ], calculating the slope K1 of the first line segment and the slope K2, K1 ═ R-G, and K2 ═ B-G of the second line segment, obtaining the slope array of each pixel point relative to the color channel, and obtaining the final slope array [ [ GN-RN, BN-GN ], [ GN-RN, BN-GN ] … ] which is relative to the color signal change condition and is marked as [ [ slopa, slopb ], [ slopa, slopb ] …;
D. performing a first round of aggregation, comparing the change trends of the RGB signals in the neighborhood of the pixel points, and dividing the slopes in the same direction in the slope array into the same type, namely judging the slopes to belong to the same region;
E. performing a second round of aggregation, calculating and comparing the similarity of RGB color signals between the regions defined in the first round, judging whether two regions with certain similarity are combined according to a preset similarity threshold value, and combining the regions which can be combined;
F. and outputting the final combined region judgment result.
2. The method as claimed in claim 1, wherein the step E specifically includes:
E1. mapping the [ R, G, B ] signals of the single pixels of each region to a one-dimensional space to obtain corresponding one-dimensional values, and combining the one-dimensional values of the single pixels of each region to form a one-dimensional array of the region;
E2. respectively sequencing and sampling the one-dimensional arrays of every two adjacent regions, calculating the similarity between the two arrays of every two adjacent regions and obtaining the maximum similarity between all the regions;
E3. multiplying the maximum similarity by the distance factor to obtain a final similarity; the distance factor a is (m-n)/m, wherein m is the maximum distance of the image, and a is the minimum distance between the two areas;
E4. and comparing whether the final similarity exceeds a preset similarity threshold, if so, judging that the two areas can be merged, otherwise, not merging.
3. The method as claimed in claim 2, wherein the step E2 of sorting and sampling the one-dimensional arrays includes sorting the values in the arrays in descending order, comparing the numbers of the numbers included in the two arrays, and keeping the first n numbers for the array with the larger number of numbers, where n is the number of the numbers included in the array with the smaller number of numbers.
4. The method as claimed in claim 3, wherein the step E2 specifically uses a cosine similarity calculation method to calculate the similarity between groups, and the calculation formula is as follows:
wherein, XiOne-dimensional value, Y, representing each pixel of one of two compared arraysiRepresenting a one-dimensional value of each pixel of another array.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910887174.7A CN110619648B (en) | 2019-09-19 | 2019-09-19 | Method for dividing image area based on RGB change trend |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910887174.7A CN110619648B (en) | 2019-09-19 | 2019-09-19 | Method for dividing image area based on RGB change trend |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110619648A true CN110619648A (en) | 2019-12-27 |
| CN110619648B CN110619648B (en) | 2022-03-15 |
Family
ID=68923726
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910887174.7A Active CN110619648B (en) | 2019-09-19 | 2019-09-19 | Method for dividing image area based on RGB change trend |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110619648B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111229649A (en) * | 2020-01-20 | 2020-06-05 | 湖北民族大学 | Article grading method, article grading equipment and automatic article grading system |
| CN112381812A (en) * | 2020-11-20 | 2021-02-19 | 深圳市优象计算技术有限公司 | Simple and efficient image quality evaluation method and system |
| CN114078139A (en) * | 2021-11-25 | 2022-02-22 | 四川长虹电器股份有限公司 | Image post-processing method based on portrait segmentation model generation result |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103366361A (en) * | 2012-04-05 | 2013-10-23 | 联发科技（新加坡）私人有限公司 | Region growing method and region growing method having marking function |
| CN103679760A (en) * | 2013-12-05 | 2014-03-26 | 河海大学 | Color image segmentation method based on Normalized cut |
| CN105184763A (en) * | 2014-06-23 | 2015-12-23 | 索尼公司 | Image processing method and device |
| CN105574841A (en) * | 2014-11-05 | 2016-05-11 | Tcl集团股份有限公司 | Color-based image segmenting method and device |
| CN106097313A (en) * | 2016-06-02 | 2016-11-09 | 甘肃读者动漫科技有限公司 | Image partition method and device |
| WO2017088365A1 (en) * | 2015-11-26 | 2017-06-01 | 乐视控股（北京）有限公司 | Skin-colour detection method and apparatus |
| CN107301405A (en) * | 2017-07-04 | 2017-10-27 | 上海应用技术大学 | Method for traffic sign detection under natural scene |
| CN107862698A (en) * | 2017-11-29 | 2018-03-30 | 首都师范大学 | Light field foreground segmentation method and device based on K mean cluster |
| CN108012131A (en) * | 2017-11-30 | 2018-05-08 | 四川长虹电器股份有限公司 | Projector image edge blending system and method |
| CN108961265A (en) * | 2018-05-30 | 2018-12-07 | 南京汇川图像视觉技术有限公司 | A kind of precision target dividing method based on color conspicuousness and Gauss model |
-
2019
- 2019-09-19 CN CN201910887174.7A patent/CN110619648B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103366361A (en) * | 2012-04-05 | 2013-10-23 | 联发科技（新加坡）私人有限公司 | Region growing method and region growing method having marking function |
| CN103679760A (en) * | 2013-12-05 | 2014-03-26 | 河海大学 | Color image segmentation method based on Normalized cut |
| CN105184763A (en) * | 2014-06-23 | 2015-12-23 | 索尼公司 | Image processing method and device |
| CN105574841A (en) * | 2014-11-05 | 2016-05-11 | Tcl集团股份有限公司 | Color-based image segmenting method and device |
| WO2017088365A1 (en) * | 2015-11-26 | 2017-06-01 | 乐视控股（北京）有限公司 | Skin-colour detection method and apparatus |
| CN106097313A (en) * | 2016-06-02 | 2016-11-09 | 甘肃读者动漫科技有限公司 | Image partition method and device |
| CN107301405A (en) * | 2017-07-04 | 2017-10-27 | 上海应用技术大学 | Method for traffic sign detection under natural scene |
| CN107862698A (en) * | 2017-11-29 | 2018-03-30 | 首都师范大学 | Light field foreground segmentation method and device based on K mean cluster |
| CN108012131A (en) * | 2017-11-30 | 2018-05-08 | 四川长虹电器股份有限公司 | Projector image edge blending system and method |
| CN108961265A (en) * | 2018-05-30 | 2018-12-07 | 南京汇川图像视觉技术有限公司 | A kind of precision target dividing method based on color conspicuousness and Gauss model |
Non-Patent Citations (5)
| Title |
|---|
| ALDO V. WANGENHEIM等: "Color image segmentation guided by a color gradient network", 《PATTERN RECOGNITION LETTERS》 * |
| SANMATI S. KAMATH等: "COLOR IMAGE SEGMENTATION IN RGB USING VECTOR ANGLE AND ABSOLUTE DIFFERENCE MEASURES", 《14TH EUROPEAN SIGNAL PROCESSING CONFERENCE》 * |
| 李艳文等: "一种基于RGB颜色分量的枝干图像分割方法", 《燕山大学学报》 * |
| 段汕等: "基于切线斜率变化统计的红色圆形交通标志牌检测", 《中南民族大学学报（自然科学版）》 * |
| 王立群等: "基于颜色恒常性的彩色图像分割方法", 《东北电力大学学报》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111229649A (en) * | 2020-01-20 | 2020-06-05 | 湖北民族大学 | Article grading method, article grading equipment and automatic article grading system |
| CN111229649B (en) * | 2020-01-20 | 2023-09-01 | 湖北民族大学 | Article grading method and automatic grading system |
| CN112381812A (en) * | 2020-11-20 | 2021-02-19 | 深圳市优象计算技术有限公司 | Simple and efficient image quality evaluation method and system |
| CN114078139A (en) * | 2021-11-25 | 2022-02-22 | 四川长虹电器股份有限公司 | Image post-processing method based on portrait segmentation model generation result |
| CN114078139B (en) * | 2021-11-25 | 2024-04-16 | 四川长虹电器股份有限公司 | Image post-processing method based on human image segmentation model generation result |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110619648B (en) | 2022-03-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110619648B (en) | Method for dividing image area based on RGB change trend | |
| CN109636784B (en) | Image saliency target detection method based on maximum neighborhood and super-pixel segmentation | |
| JP5008572B2 (en) | Image processing method, image processing apparatus, and computer-readable medium | |
| CN110866896B (en) | Image saliency target detection method based on k-means and level set super-pixel segmentation | |
| CN108564085B (en) | Method for automatically reading of pointer type instrument | |
| CN107038416B (en) | Pedestrian detection method based on binary image improved HOG characteristics | |
| CN112991238B (en) | Food image segmentation method, system and medium based on texture and color mixing | |
| Renugambal et al. | Application of image processing techniques in plant disease recognition | |
| CN108320289B (en) | Bone marrow cell segmentation method based on sparse representation and morphological operation | |
| Niu et al. | Image segmentation algorithm for disease detection of wheat leaves | |
| CN108921820A (en) | A kind of saliency object detection method based on feature clustering and color contrast | |
| Ani Brown Mary et al. | Classification of coral reef submarine images and videos using a novel Z with tilted Z local binary pattern (Z⊕ TZLBP) | |
| CN110070545B (en) | Method for automatically extracting urban built-up area by urban texture feature density | |
| CN112132166A (en) | Intelligent analysis method, system and device for digital cytopathology image | |
| US8000535B2 (en) | Methods and systems for refining text segmentation results | |
| CN105320970A (en) | Potato disease diagnostic device, diagnostic system and diagnostic method | |
| CN103678552A (en) | Remote-sensing image retrieving method and system based on salient regional features | |
| Reta et al. | Color uniformity descriptor: An efficient contextual color representation for image indexing and retrieval | |
| CN115033721A (en) | Image retrieval method based on big data | |
| CN109635715B (en) | Remote sensing image building extraction method | |
| Senturk et al. | Seam carving based image retargeting: A survey | |
| Jamil et al. | A rule-based segmentation method for fruit images under natural illumination | |
| CN112184696A (en) | Method and system for counting cell nucleus and cell organelle and calculating area of cell nucleus and cell organelle | |
| CN110490210B (en) | Color texture classification method based on t sampling difference between compact channels | |
| CN105574880A (en) | Color image segmentation method based on exponential moment pixel classification |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |