CN104966085A - Remote sensing image region-of-interest detection method based on multi-significant-feature fusion - Google Patents
Remote sensing image region-of-interest detection method based on multi-significant-feature fusion Download PDFInfo
- Publication number
- CN104966085A CN104966085A CN201510331174.0A CN201510331174A CN104966085A CN 104966085 A CN104966085 A CN 104966085A CN 201510331174 A CN201510331174 A CN 201510331174A CN 104966085 A CN104966085 A CN 104966085A
- Authority
- CN
- China
- Prior art keywords
- mrow
- color
- image
- remote sensing
- color channel
- 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
- 230000004927 fusion Effects 0.000 title claims abstract description 8
- 238000001514 detection method Methods 0.000 title abstract description 22
- 238000000034 method Methods 0.000 claims abstract description 52
- 230000011218 segmentation Effects 0.000 claims abstract description 5
- 238000012545 processing Methods 0.000 claims abstract description 4
- 238000004364 calculation method Methods 0.000 claims description 11
- 238000010586 diagram Methods 0.000 claims description 9
- 238000003064 k means clustering Methods 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 8
- 238000001914 filtration Methods 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 abstract description 2
- 238000011835 investigation Methods 0.000 abstract 1
- 230000000007 visual effect Effects 0.000 description 19
- 238000004458 analytical method Methods 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 3
- 238000009499 grossing Methods 0.000 description 3
- 101150112439 SMAP gene Proteins 0.000 description 2
- 230000000052 comparative effect Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000004456 color vision Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000004424 eye movement Effects 0.000 description 1
- 230000004438 eyesight Effects 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Classifications
-
- 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]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a remote sensing image region-of-interest detection method based on multi-significant-feature fusion, belonging to the technical fields of remote sensing image processing and image identification. The remote sensing image region-of-interest detection method comprises the following steps: 1) obtaining color channels of one group of input remote sensing images and calculating a color histogram of each color channel; 2) calculating a standard significant weight of each color channel according to the color histograms; 3) calculating an information content significant feature image; 4) converting one group of input remote sensing images from an RGB color space to a CIE Lab color space; 5) utilizing a clustering algorithm to obtain clusters; 6) calculating a significant value of each cluster, and obtaining a common significant feature image; 7) fusing the information content significant feature image with the common significant feature image to obtain a final significant image; and 8) performing threshold segmentation through an OTSU method to extract a region of interest. Compared with a traditional method, the remote sensing image region-of-interest detection method of the present invention achieves accurate detection for a remote sensing image region-of-interest on the premise of not having a prior knowledge base, thus the remote sensing image region-of-interest detection method can be widely applied to fields such as environment monitoring, land utilization and agricultural investigation.
Description
Technical Field
The invention belongs to the technical field of remote sensing image processing and image recognition, and particularly relates to a remote sensing image region-of-interest detection method based on multi-salient feature fusion.
Background
With the rapid development of remote sensing technology, the data scale of remote sensing images is rapidly enlarged, and the extraction of the region of interest of the remote sensing images can reduce the complexity of analysis and processing of the remote sensing images, so that the extraction of the region of interest of the remote sensing images is also a focus of attention for a period of time recently, and how to accurately and rapidly realize the detection of the region of interest of the remote sensing images becomes one of the problems to be solved urgently. The problem can be effectively solved, the contradiction between high-speed acquisition and low-speed interpretation of the remote sensing image is relieved, and the method has important practical application value in the relevant fields of land utilization, disaster assessment, town planning, environment monitoring and the like.
The detection of the region of interest of the traditional remote sensing image is mostly based on the global situation and needs prior knowledge. However, the establishment of the prior knowledge base is a very complicated problem, and the information such as the expert knowledge base, the characteristics of the target area, the characteristics of the background area and the like needs to be comprehensively considered. Some methods need to introduce training of psychophysical data of color presentation and eye movement, and some methods need to detect and classify the remote sensing image interesting region by means of a digital map of the same region. These algorithms all require a priori knowledge base and have high computational complexity.
The visual attention model provides a new visual angle for the remote sensing image interesting area detection, is different from the traditional detection method, is completely driven by data, does not relate to the influence of external factors such as a knowledge base and the like, has the advantages of quick identification, accurate result and the like, is paid more and more attention, and has great significance for introducing the visual attention model into the remote sensing image interesting area detection.
In terms of a visual Attention Model based on low-level visual features, the Itti et al, in the article "A Model of personal-basic visual Attention for Rapid Scene Analysis", proposes an Itti visual Attention method, which approximates the human visual system and utilizes various visual characteristics to generate Saliency maps. In terms of mathematical-Based Visual attention models, Harel et al propose a Graph-Based Visual attention algorithm (GBVS) in the article "Graph-Based Visual attention", which completes the feature extraction step by simulating the Visual attention mechanism using a conventional Itti model, then represents the pixel associations between images using Graph structures, and finally introduces a Markov chain (Markow chains) to compute Saliency maps. In terms of a Frequency domain analysis attention model, Achanta et al propose a Frequency-tuned, FT, method for Salient region detection in the article "Frequency-tuned significant region detection", convert an input RGB image into a CIELab color space and perform gaussian smoothing, subtract an arithmetic average of image feature vectors, and find an amplitude by points to obtain a uniform and clear-bounded saliency map.
The visual attention model based on the low-level visual features better simulates the attention mode of human vision, but the frequency domain features of images are not fully considered, and meanwhile, the calculation speed is low, the efficiency is low, and the requirement of real-time application is difficult to achieve. The visual attention model based on the frequency domain analysis method is concise in form and easy to interpret and implement, but when the proportion of the salient region in the whole image is too large or the background of the image is too complex, the salient map obtained by the method can mark part of the background as the salient region by mistake, and the biological rationality of the salient region is not very clear. In recent years, scholars at home and abroad also propose a new algorithm for applying the visual saliency to the remote sensing image region-of-interest detection. For example, Zhang et al, in the article "Fast Detection of Visual quality Regions in remote Sensing Image based Region Growing", propose to reduce the Image resolution based on wavelet transform, introduce two-dimensional discrete moment transform in the Visual features, and generate a Saliency map. However, these algorithms have a common disadvantage in that they can only extract salient regions, but cannot distinguish the differences between the salient regions. And if the similarity of a group of remote sensing images with similar interested areas can be utilized, other areas interfering with the detection of the interested areas can be excluded.
In the aspect of calculating the mask of the region of interest, the conventional method usually uses a fixed radius circle to describe the region of interest, which brings a lot of redundant information when identifying random regions, but the speed of using a single threshold value is very fast, but the region of interest has many small fragments, and the region description is not accurate. The maximum inter-class variance method (Ostu method) is an automatic non-parametric and unsupervised threshold selection method, is a simple and efficient method for adaptively calculating a single threshold, and has the advantages of simplicity in calculation, strong self-adaptation and the like.
Disclosure of Invention
The invention aims to provide a remote sensing image region-of-interest detection method based on multi-salient feature fusion, which is used for accurately detecting a region-of-interest of a remote sensing image. The existing region-of-interest detection method is mainly based on the global situation and needs prior knowledge. However, the establishment of the prior knowledge base is a very complicated problem, and the information such as the expert knowledge base, the characteristics of the target area, the characteristics of the background area and the like needs to be comprehensively considered. The method of the invention therefore focuses mainly on two aspects:
1) a priori knowledge base is not required to be established based on global search;
2) the detection precision of the region of interest of the remote sensing image is improved, and more accurate region of interest information is obtained.
The technical scheme adopted by the invention comprises five main processes of information quantity significant characteristic diagram generation of the remote sensing image, common significant characteristic diagram generation, final significant diagram generation, interested region template generation and interested region generation, and specifically comprises the following steps:
the method comprises the following steps: calculating color histogram, i.e. inputting a group of remote sensing images with size of M × N, respectively extracting each color channel of each image, and using fc(x, y) represents the color intensity of the (x, y) position in the color channel c, and an intensity histogram H of each remote sensing image in different color channels is constructedc(i) Where M denotes the length of the image, N denotes the width of the image, x and y denote the abscissa and ordinate of the image, respectively, x is 1, 2 … … M, y is 1, 2 … … N, c denotes the color channel, c is 1, 2, 3, i denotes the pixel intensity value, i is 0, 1 ……255；
Step two: calculating a normalized saliency weight for a color channel c, i.e. from a color histogram H of the color channel cc(i) Calculating the information amount In of each pixel intensity value i In the color channelc(i) Assigning the information quantity to a pixel point with the same intensity value as the pixel point, and obtaining an information quantity graph LOG of the color channel c after finishing all calculation and assignmentc(x, y) obtaining the significance h of the color channel c by using the information quantity diagramcAnd then calculating to obtain the standardized significance weight w of each color channel of each image by using the significance of each color channelc；
Step three: calculating saliency maps of information quantities, i.e. using normalized saliency weights w for each color channelcWeighting calculation is carried out to obtain a preliminary information content significant feature map of each image, Gaussian smooth filtering is carried out on the preliminarily obtained information content significant feature map, and noise is filtered to obtain a final information content significant feature map of each image;
step four: converting a group of remote sensing images from an RGB color space to a CIE Lab color space, namely respectively extracting R, G, B three color channel values of each pixel of each image, converting the three color channel values into the CIE Lab color space, and acquiring three components of L, a and B, wherein in the RGB color space, R represents red, G represents green, B represents blue, and in the CIELab color space, L represents brightness, L is 0 represents black, L is 100 represents white, a represents the position of the color between red/green, a is a negative value and represents green, a is a positive value and represents red, B represents the position of the color between blue/yellow, B is a negative value and represents blue, and B is a positive value and represents yellow;
step five: finishing pixel clustering of a CIE Lab color space by using a k-means clustering algorithm, namely mapping the group of original remote sensing images to values of all pixel points on the CIE Lab color space for clustering by using the k-means clustering algorithm to obtain k clusters;
step six: calculating a common significant feature map, namely dividing the number of pixels contained in the jth cluster by the total number of pixels of the image, wherein the division result is defined as the weight of the jth cluster, wherein j is 1 and 2 … … k, calculating the significant value of the cluster by using the weight of the cluster and the distance between the clusters, and assigning the significant value of the cluster to each pixel point belonging to the cluster, thereby obtaining a group of common significant feature maps;
step seven: calculating a final saliency map, namely multiplying an information quantity saliency characteristic map obtained by utilizing histogram information of each color channel by a common saliency characteristic map obtained by k-means clustering in a CIE Lab color space, so as to obtain the final saliency map after fusion of multiple saliency characteristics;
step eight: and extracting the region of interest, namely obtaining a segmentation threshold of the final saliency map by a maximum inter-class variance method, segmenting the final saliency map into a binary image template by using the threshold, representing the region of interest by using '1', representing a region of non-interest by using '0', and finally multiplying the binary image template and the original image to obtain a final region of interest extraction result.
Drawings
FIG. 1 is a flow chart of the present invention.
Fig. 2 is an exemplary set of four remote sensing images used in the present invention.
Fig. 3 is a characteristic diagram and a final saliency map of the present invention. (a) The information content saliency map of the example picture, (b) the common saliency map of the example picture, and (c) the final saliency map of the example picture.
Fig. 4 is a comparison of saliency maps generated using the method of the present invention and other methods for an example picture. (a) A saliency map generated for the Itti method, (b) a saliency map generated for the GBVS method, (c) a saliency map generated for the FT method, and (d) a saliency map generated by the method of the invention.
Fig. 5 is a comparison of regions of interest detected in exemplary pictures using the method of the present invention and other methods. (a) A map of regions of interest detected for the Itti method, (b) a region of interest detected for the GBVS method, (c) a region of interest detected for the FT method, and (d) a region of interest detected for the method of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings. The general framework of the invention is shown in fig. 1, and details of each step of the implementation will now be described.
The method comprises the following steps: calculating a color histogram;
inputting a group of remote sensing images with size of M multiplied by NAs shown in FIG. 2, each image I is obtained separatelypFor each color channel of (1), with fc(x, y) denotes the image IpConstructing an intensity histogram H of the remote sensing image in different color channels according to the color intensity of (x, y) positions in the color channel cc(i) Wherein M represents the length of the image, N represents the width of the image, and the total number of the group of remote sensing images is QRepresenting a number Q of groups of remote sensing images, IpRepresenting the pth of a group of remote sensing images, wherein p is 1 and 2 … … Q, x and y respectively represent the abscissa and ordinate of the image, x is 1 and 2 … … M, y is 1 and 2 … … N, c represents a color channel, c is 1, 2 and 3, i represents a pixel intensity value, and i is 0 and 1 … … 255 5;
the histogram for each color channel of each image in the set of images may be obtained using the following formula:
wherein,c(x, y) represents the binarized image of color channel c, and the calculation formula is:
step two: calculating a normalized saliency weight for color channel c;
according to image IpColor histogram H of color channel cc(i) Calculating each of the color channelsInformation quantity In of individual pixel intensity values ic(i) Calculating and assigning by using the information quantity to finally obtain an image IpNormalized saliency weight w for each color channel ofcThe method is realized by the following four steps;
(1) according to image IpColor histogram H in color channel c ofc(i) The information amount In of each pixel intensity value In the color channel is calculated by using the following formulac(i)：
In(i)c＝-ln(Hc(i))
(2) Assigning the information quantity to a pixel point in the color channel c with the same intensity value as the pixel point to obtain an information quantity graph LOG of the color channel cc(x, y), namely:
i＝fc(x，y)
(3) information quantity map LOG using color channel cc(x, y) calculating to obtain the significance hcThe calculation formula is as follows:
where there are three color channels, then h1Signifying the saliency of color channel 1, h2Representing the saliency, h, of the color channel 23Represents the saliency of the color channel 3;
(4) dividing the significance of the color channel c by the significance of the three color channels to obtain a negative logarithm of the number to obtain the color channel standardized significance weight wc：
Where there are three color channels, then w1To representNormalized saliency weight, w, for color channel 12Normalized saliency weight, w, representing color channel 23A normalized saliency weight representing a color channel 3;
step three: calculating an information quantity significant feature map;
using images IpNormalized saliency weight w of each color channel ofcWeighting calculation to obtain a preliminary information content significant feature map Smap (x, y), performing Gaussian smoothing filtering on the preliminary information content significant feature map, and filtering noise to obtain a final information content significant feature map SS (x, y):
wherein,representing a gaussian smoothing filter;
through the steps, the remote sensing image group is obtainedThe information content of each remote sensing image in the image is a salient feature map.
Step four: converting the remote sensing image from the RGB color space to the CIE Lab color space;
because the color channel of the CIELab removes the brightness information to a certain extent, the reflected content is closer to the essence of color perception, so that the color smoothness can be better embodied, and based on the obvious advantages of the CIE Lab space in color uniformity, clustering is carried out on the CIE Lab color space, and the following advanced color space conversion is carried out:
respectively extracting remote sensing image groupConverting R, G, B three color channel values of each pixel of each image into CIE Lab color space, obtaining three components of L, a and b of each pixel, and recording the remote sensing image group in the CIE Lab color space asIn the RGB color space, R represents red, G represents green, B represents blue, three channels of the CIE Lab color space represent brightness L, L0 represents black, L100 represents white, the color is at a position a between red/green, a is green with a negative value, a is red with a positive value, the color is at a position B between blue/yellow, B is blue with a negative value, and B is yellow with a positive value;
step five: clustering color features;
the method comprises the following steps of finishing pixel clustering of a CIE Lab color space by utilizing a k-means clustering algorithm, namely clustering values of all pixel points of a group of images in the CIE Lab color space to obtain k clusters, wherein the specific implementation steps are as follows:
(1) extracting remote sensing image groupIn three channels L, a and b of CIE Lab color space, adjusting the range of pixel point values in the three channels to make the adjusted range of pixel point values in the three channelsThe same;
(2) and simultaneously calculating pixel values of three channels of all the images in the image group to ensure that the sum of squared distances between each pixel point value and the nearest clustering center is minimum, wherein at the moment, all pixel points with the same nearest clustering center are a cluster, and the sum of squared distances W can be calculated by utilizing the following formula:
in the formula pirRepresenting pixel value, where r is 1, 2 … … n, n is the number of image pixel points, ajRepresents a cluster center, where j ═ 1, 2 … … k;
step six: calculating a common salient feature map;
after the weights of all k clusters are obtained through calculation, the significant value of the cluster can be calculated by using the weights of the clusters and the distance between the clusters, and the significant value of the cluster is assigned to each pixel point belonging to the cluster, so that a group of common significant feature graphs are obtained, and the specific implementation needs the following three steps:
(1) the jth cluster ljThe number of pixels contained in (1) is divided by the total number of pixels in the image group, and the division result is defined as the weight ω (l) of the jth clusterj) Wherein j is 1, 2 … … k;
(2) definition D (l)t，lj) Is divided into two clusters lt、ljA color distance of eachOne cluster of significant values CL (l)j) Can be calculated by the following formula:
wherein,
wherein j and t are 1, 2 … … k and qtsIs the probability that the s-th color appears in the m colors of the t-th cluster, i.e., there are m pixel values in the t-th cluster, and s is 1, 2 … … m;
(3) clustering to enable the significance value of each pixel point to be equal to the significance value of the cluster where the pixel point is located, and thus obtaining a common characteristic significance map SM (x, y):
when ILabp(x，y)∈ljWherein j is 1, 2 … … k, p is 1, 2 … … Q,
SM(x，y)＝CL(lj)
through the steps, the remote sensing image group is obtainedA common saliency map of each of the remote sensing images.
Step seven: calculating a final saliency map;
correspondingly multiplying the information quantity significant feature diagram obtained through each color channel with a common significant feature diagram obtained through k-means clustering in a CIE Lab color space, thereby obtaining a final significant diagram S (x, y) after the multiple significant features of each image in the group of remote sensing images are fused:
S(x，y)＝SS(x，y)×SM(x，y)
step eight: extracting an interested region;
obtaining a segmentation threshold of the final saliency map by a maximum inter-class variance method, segmenting the final saliency map into a binary image template by using the threshold, representing an interested region by using '1', representing a non-interested region by using '0', and finally multiplying the binary image template and the original image to obtain a final interested region extraction result.
The effects of the present invention can be further illustrated by the following experimental results and analyses:
1. experimental data
The invention selects a group of visible light remote sensing images of a suburb of Beijing from a SPOT5 satellite source image, and respectively intercepts and generates a group of images with the size of 1024 multiplied by 1024 as a source image of the experiment, as shown in figure 2.
2. Comparative experiment
In order to evaluate the performance of the method, the following comparative experiment is designed, and the ITTI method, the GBVS method and the FT method are selected for comparing the performance with the method disclosed by the invention. The saliency map and the region of interest map generated by the different methods were subjectively compared, as shown in fig. 4 and 5, respectively. In fig. 4, (a) is a saliency map generated by the Itti method, (b) is a saliency map generated by the GBVS method, (c) is a saliency map generated by the FT method, and (d) is a saliency map generated by the method of the present invention. In fig. 5, (a) is the region of interest map generated by the Itti method, (b) is the region of interest map generated by the GBVS method, (c) is the region of interest map generated by the FT method, and (d) is the region of interest map generated by the method of the present invention.
By contrast, the resolution of the saliency map obtained using the iti model is low, only 1/256 of the size of the original map, and the saliency map is enlarged when the region of interest is finally extracted. The GBVS model is based on the Itti model, and only utilizes a Markov chain when a saliency map is obtained. The regions of interest obtained from the two models are larger than the regions that need to be extracted originally, i.e. unnecessary parts are extracted. By using the FT model, a better extraction result can be obtained when the background frequency changes little, but the extraction result is interfered when the background frequency changes greatly, and the algorithm can obtain a better detection result.
Claims (2)
1. A method for detecting a region of interest of a remote sensing image based on multi-salient feature fusion is characterized by processing a group of remote sensing images, firstly utilizing color information of the remote sensing images, obtaining an information quantity salient feature map by constructing color histograms of different color channels and carrying out weighted calculation, secondly utilizing a k-means clustering algorithm to cluster the group of remote sensing images on a CIE Lab color space and calculating a salient value, thereby obtaining a group of common salient feature maps of the CIE Lab color space, then fusing the two groups of maps to obtain a final salient map, and finally carrying out threshold segmentation by a maximum inter-class variance method to extract the region of interest, and the method is characterized by comprising the following steps of:
the method comprises the following steps: calculating color histogram, i.e. inputting a group of remote sensing images with size of M × N, respectively extracting each color channel of each image, and using fc(x, y) represents the color intensity of the (x, y) position in the color channel c, and an intensity histogram H of each remote sensing image in different color channels is constructedc(i) Where M denotes the length of the image, N denotes the width of the image, x and y denote the abscissa and ordinate of the image, respectively, x is 1, 2 … … M, y is 1, 2 … … N, c denotes the color channel, c is 1, 2, 3, i denotes the pixel intensity value, i is 0, 1 … … 255;
step two: calculating a normalized saliency weight for a color channel c, i.e. from a color histogram H of the color channel cc(i) Calculating the information amount In of each pixel intensity value i In the color channelc(i) Assigning the information quantity to a pixel point with the same intensity value as the pixel point, and obtaining an information quantity graph LOG of the color channel c after finishing all calculation and assignmentc(x, y) obtaining the significance h of the color channel c by using the information quantity diagramcAnd then calculating to obtain the standardized significance weight w of each color channel of each image by using the significance of each color channelc；
Step three: calculating saliency maps of information quantities, i.e. using normalized saliency weights w for each color channelcWeighting calculation is carried out to obtain a preliminary information content significant feature map of each image, Gaussian smooth filtering is carried out on the preliminarily obtained information content significant feature map, and noise is filtered to obtain a final information content significant feature map of each image;
step four: converting a group of remote sensing images from an RGB color space to a CIE Lab color space, namely respectively extracting R, G, B three color channel values of each pixel of each image, converting the three color channel values into the CIE Lab color space, and acquiring three components of L, a and B, wherein in the RGB color space, R represents red, G represents green, B represents blue, and in the CIE Lab color space, L represents brightness, L is 0 represents black, L is 100 represents white, a represents the position of the color between red/green, a is a negative value and represents green, a is a positive value and represents red, B represents the position of the color between blue/yellow, B is a negative value and represents blue, and B is a positive value and represents yellow;
step five: finishing pixel clustering of a CIE Lab color space by using a k-means clustering algorithm, namely mapping the group of original remote sensing images to values of all pixel points on the CIE Lab color space for clustering by using the k-means clustering algorithm to obtain k clusters;
step six: calculating a common significant feature map, namely dividing the number of pixels contained in the jth cluster by the total number of pixels of the image, wherein the division result is defined as the weight of the jth cluster, wherein j is 1 and 2 … … k, calculating the significant value of the cluster by using the weight of the cluster and the distance between the clusters, and assigning the significant value of the cluster to each pixel point belonging to the cluster, thereby obtaining a group of common significant feature maps;
step seven: calculating a final saliency map, namely multiplying an information quantity saliency characteristic map obtained by utilizing histogram information of each color channel by a common saliency characteristic map obtained by k-means clustering in a CIE Lab color space, so as to obtain the final saliency map after fusion of multiple saliency characteristics;
step eight: and extracting the region of interest, namely obtaining a segmentation threshold of the final saliency map by a maximum inter-class variance method, segmenting the final saliency map into a binary image template by using the threshold, representing the region of interest by using '1', representing a region of non-interest by using '0', and finally multiplying the binary image template and the original image to obtain a final region of interest extraction result.
2. The method for extracting the region of interest of the remote sensing image based on the salient feature clustering is characterized in that the specific process of the second step is as follows:
1) according to the color histogram H in the color channel cc(i) Calculating the information amount In of each pixel intensity valuec(i)：
In(i)c＝-ln(Hc(i))
2) Assigning the information quantity to a pixel point with the same intensity value as the pixel point to obtain an information quantity graph LOG of a color channel cc(x, y), namely:
i＝fc(x，y)，
3) information quantity map LOG using color channel cc(x, y) calculating to obtain the significance hcSince the image contains three color channels, h is used1Signifying the saliency of color channel 1, h2Representing the saliency, h, of the color channel 23Significance of the color channel 3:
4) dividing the significance of the color channel c by the sum of the significance of the three color channels to obtain a negative logarithm of the number, and obtaining the color channel standardized significance weight wc，
Since the image contains three color channels, w is used1Normalized saliency weight, w, representing color channel 12Normalized saliency weight, w, representing color channel 23Representing the normalized saliency weight of the color channel 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510331174.0A CN104966085B (en) | 2015-06-16 | 2015-06-16 | A kind of remote sensing images region of interest area detecting method based on the fusion of more notable features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510331174.0A CN104966085B (en) | 2015-06-16 | 2015-06-16 | A kind of remote sensing images region of interest area detecting method based on the fusion of more notable features |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104966085A true CN104966085A (en) | 2015-10-07 |
| CN104966085B CN104966085B (en) | 2018-04-03 |
Family
ID=54220120
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510331174.0A Expired - Fee Related CN104966085B (en) | 2015-06-16 | 2015-06-16 | A kind of remote sensing images region of interest area detecting method based on the fusion of more notable features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104966085B (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407978A (en) * | 2016-09-24 | 2017-02-15 | 上海大学 | Unconstrained in-video salient object detection method combined with objectness degree |
| CN106780422A (en) * | 2016-12-28 | 2017-05-31 | 深圳市美好幸福生活安全系统有限公司 | A kind of notable figure fusion method based on Choquet integrations |
| CN106951841A (en) * | 2017-03-09 | 2017-07-14 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | A kind of multi-object tracking method based on color and apart from cluster |
| CN107239760A (en) * | 2017-06-05 | 2017-10-10 | 中国人民解放军军事医学科学院基础医学研究所 | A kind of video data handling procedure and system |
| CN108335307A (en) * | 2018-04-19 | 2018-07-27 | 云南佳叶现代农业发展有限公司 | Adaptive tobacco leaf picture segmentation method and system based on dark primary |
| CN108364288A (en) * | 2018-03-01 | 2018-08-03 | 北京航空航天大学 | Dividing method and device for breast cancer pathological image |
| CN108596920A (en) * | 2018-05-02 | 2018-09-28 | 北京环境特性研究所 | A kind of Target Segmentation method and device based on coloured image |
| CN108764106A (en) * | 2018-05-22 | 2018-11-06 | 中国计量大学 | Multiple dimensioned colour image human face comparison method based on cascade structure |
| CN109035254A (en) * | 2018-09-11 | 2018-12-18 | 中国水产科学研究院渔业机械仪器研究所 | Based on the movement fish body shadow removal and image partition method for improving K-means cluster |
| CN109858394A (en) * | 2019-01-11 | 2019-06-07 | 西安电子科技大学 | A kind of remote sensing images water area extracting method based on conspicuousness detection |
| CN109949906A (en) * | 2019-03-22 | 2019-06-28 | 上海鹰瞳医疗科技有限公司 | Pathological section image procossing and model training method and equipment |
| CN110232378A (en) * | 2019-05-30 | 2019-09-13 | 苏宁易购集团股份有限公司 | A kind of image interest point detecting method, system and readable storage medium storing program for executing |
| CN110268442A (en) * | 2019-05-09 | 2019-09-20 | 京东方科技集团股份有限公司 | In the picture detect background objects on exotic computer implemented method, in the picture detect background objects on exotic equipment and computer program product |
| CN110612534A (en) * | 2017-06-07 | 2019-12-24 | 赫尔实验室有限公司 | System for detecting salient objects in images |
| CN111339953A (en) * | 2020-02-27 | 2020-06-26 | 广西大学 | Clustering analysis-based mikania micrantha monitoring method |
| CN111400557A (en) * | 2020-03-06 | 2020-07-10 | 北京市环境保护监测中心 | Method and device for automatically identifying atmospheric pollution key area |
| CN113139934A (en) * | 2021-03-26 | 2021-07-20 | 上海师范大学 | Rice grain counting method |
| CN113469976A (en) * | 2021-07-06 | 2021-10-01 | 浙江大华技术股份有限公司 | Object detection method and device and electronic equipment |
| CN115131327A (en) * | 2022-07-14 | 2022-09-30 | 电子科技大学 | Color feature fused display screen color line defect detection method |
| CN118570201A (en) * | 2024-08-01 | 2024-08-30 | 吴江市兰天织造有限公司 | Ultra-high density fabric detection method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100239170A1 (en) * | 2009-03-18 | 2010-09-23 | Asnis Gary I | System and method for target separation of closely spaced targets in automatic target recognition |
| US20120051606A1 (en) * | 2010-08-24 | 2012-03-01 | Siemens Information Systems Ltd. | Automated System for Anatomical Vessel Characteristic Determination |
| CN103810710A (en) * | 2014-02-26 | 2014-05-21 | 西安电子科技大学 | Multispectral image change detection method based on semi-supervised dimensionality reduction and saliency map |
| CN104463224A (en) * | 2014-12-24 | 2015-03-25 | 武汉大学 | Hyperspectral image demixing method and system based on abundance significance analysis |
-
2015
- 2015-06-16 CN CN201510331174.0A patent/CN104966085B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100239170A1 (en) * | 2009-03-18 | 2010-09-23 | Asnis Gary I | System and method for target separation of closely spaced targets in automatic target recognition |
| US20120051606A1 (en) * | 2010-08-24 | 2012-03-01 | Siemens Information Systems Ltd. | Automated System for Anatomical Vessel Characteristic Determination |
| CN103810710A (en) * | 2014-02-26 | 2014-05-21 | 西安电子科技大学 | Multispectral image change detection method based on semi-supervised dimensionality reduction and saliency map |
| CN104463224A (en) * | 2014-12-24 | 2015-03-25 | 武汉大学 | Hyperspectral image demixing method and system based on abundance significance analysis |
Cited By (31)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407978A (en) * | 2016-09-24 | 2017-02-15 | 上海大学 | Unconstrained in-video salient object detection method combined with objectness degree |
| CN106780422A (en) * | 2016-12-28 | 2017-05-31 | 深圳市美好幸福生活安全系统有限公司 | A kind of notable figure fusion method based on Choquet integrations |
| CN106951841A (en) * | 2017-03-09 | 2017-07-14 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | A kind of multi-object tracking method based on color and apart from cluster |
| CN106951841B (en) * | 2017-03-09 | 2020-05-12 | 广东顺德中山大学卡内基梅隆大学国际联合研究院 | Multi-target tracking method based on color and distance clustering |
| CN107239760A (en) * | 2017-06-05 | 2017-10-10 | 中国人民解放军军事医学科学院基础医学研究所 | A kind of video data handling procedure and system |
| CN107239760B (en) * | 2017-06-05 | 2020-07-17 | 中国人民解放军军事医学科学院基础医学研究所 | Video data processing method and system |
| CN110612534A (en) * | 2017-06-07 | 2019-12-24 | 赫尔实验室有限公司 | System for detecting salient objects in images |
| CN110612534B (en) * | 2017-06-07 | 2023-02-21 | 赫尔实验室有限公司 | System, computer-readable medium, and method for detecting salient objects in an image |
| CN108364288B (en) * | 2018-03-01 | 2022-04-05 | 北京航空航天大学 | Segmentation method and device for breast cancer pathological image |
| CN108364288A (en) * | 2018-03-01 | 2018-08-03 | 北京航空航天大学 | Dividing method and device for breast cancer pathological image |
| CN108335307A (en) * | 2018-04-19 | 2018-07-27 | 云南佳叶现代农业发展有限公司 | Adaptive tobacco leaf picture segmentation method and system based on dark primary |
| CN108596920A (en) * | 2018-05-02 | 2018-09-28 | 北京环境特性研究所 | A kind of Target Segmentation method and device based on coloured image |
| CN108764106A (en) * | 2018-05-22 | 2018-11-06 | 中国计量大学 | Multiple dimensioned colour image human face comparison method based on cascade structure |
| CN108764106B (en) * | 2018-05-22 | 2021-12-21 | 中国计量大学 | Multi-scale color image face comparison method based on cascade structure |
| CN109035254A (en) * | 2018-09-11 | 2018-12-18 | 中国水产科学研究院渔业机械仪器研究所 | Based on the movement fish body shadow removal and image partition method for improving K-means cluster |
| CN109858394A (en) * | 2019-01-11 | 2019-06-07 | 西安电子科技大学 | A kind of remote sensing images water area extracting method based on conspicuousness detection |
| CN109949906A (en) * | 2019-03-22 | 2019-06-28 | 上海鹰瞳医疗科技有限公司 | Pathological section image procossing and model training method and equipment |
| CN110268442B (en) * | 2019-05-09 | 2023-08-29 | 京东方科技集团股份有限公司 | Computer-implemented method of detecting a foreign object on a background object in an image, device for detecting a foreign object on a background object in an image, and computer program product |
| CN110268442A (en) * | 2019-05-09 | 2019-09-20 | 京东方科技集团股份有限公司 | In the picture detect background objects on exotic computer implemented method, in the picture detect background objects on exotic equipment and computer program product |
| CN110232378A (en) * | 2019-05-30 | 2019-09-13 | 苏宁易购集团股份有限公司 | A kind of image interest point detecting method, system and readable storage medium storing program for executing |
| CN110232378B (en) * | 2019-05-30 | 2023-01-20 | 苏宁易购集团股份有限公司 | Image interest point detection method and system and readable storage medium |
| CN111339953A (en) * | 2020-02-27 | 2020-06-26 | 广西大学 | Clustering analysis-based mikania micrantha monitoring method |
| CN111400557A (en) * | 2020-03-06 | 2020-07-10 | 北京市环境保护监测中心 | Method and device for automatically identifying atmospheric pollution key area |
| CN111400557B (en) * | 2020-03-06 | 2023-08-08 | 北京市环境保护监测中心 | Method and device for automatically identifying important areas of atmospheric pollution |
| CN113139934A (en) * | 2021-03-26 | 2021-07-20 | 上海师范大学 | Rice grain counting method |
| CN113139934B (en) * | 2021-03-26 | 2024-04-30 | 上海师范大学 | Rice grain counting method |
| CN113469976A (en) * | 2021-07-06 | 2021-10-01 | 浙江大华技术股份有限公司 | Object detection method and device and electronic equipment |
| CN115131327A (en) * | 2022-07-14 | 2022-09-30 | 电子科技大学 | Color feature fused display screen color line defect detection method |
| CN115131327B (en) * | 2022-07-14 | 2024-04-30 | 电子科技大学 | Color line defect detection method for display screen with fused color features |
| CN118570201A (en) * | 2024-08-01 | 2024-08-30 | 吴江市兰天织造有限公司 | Ultra-high density fabric detection method |
| CN118570201B (en) * | 2024-08-01 | 2024-10-11 | 吴江市兰天织造有限公司 | Ultra-high density fabric detection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104966085B (en) | 2018-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104966085B (en) | A kind of remote sensing images region of interest area detecting method based on the fusion of more notable features | |
| CN109344701B (en) | Kinect-based dynamic gesture recognition method | |
| CN106780485B (en) | SAR image change detection method based on super-pixel segmentation and feature learning | |
| CN108288033B (en) | A kind of safety cap detection method based on random fern fusion multiple features | |
| CN110363122A (en) | A kind of cross-domain object detection method based on multilayer feature alignment | |
| CN104835175B (en) | Object detection method in a kind of nuclear environment of view-based access control model attention mechanism | |
| WO2015180527A1 (en) | Image saliency detection method | |
| CN105528595A (en) | Method for identifying and positioning power transmission line insulators in unmanned aerial vehicle aerial images | |
| CN110400293B (en) | No-reference image quality evaluation method based on deep forest classification | |
| CN108960404B (en) | Image-based crowd counting method and device | |
| CN105809121A (en) | Multi-characteristic synergic traffic sign detection and identification method | |
| CN105528794A (en) | Moving object detection method based on Gaussian mixture model and superpixel segmentation | |
| CN106909902A (en) | A kind of remote sensing target detection method based on the notable model of improved stratification | |
| CN112906550B (en) | Static gesture recognition method based on watershed transformation | |
| CN103295013A (en) | Pared area based single-image shadow detection method | |
| CN109886146B (en) | Flood information remote sensing intelligent acquisition method and device based on machine vision detection | |
| CN104392459A (en) | Infrared image segmentation method based on improved FCM (fuzzy C-means) and mean drift | |
| CN106557740A (en) | The recognition methods of oil depot target in a kind of remote sensing images | |
| CN114387505A (en) | Hyperspectral and laser radar multi-modal remote sensing data classification method and system | |
| Wang et al. | Haze removal algorithm based on single-images with chromatic properties | |
| WO2020119624A1 (en) | Class-sensitive edge detection method based on deep learning | |
| Ju et al. | A novel fully convolutional network based on marker-controlled watershed segmentation algorithm for industrial soot robot target segmentation | |
| Li et al. | Superpixel-based adaptive salient region analysis for infrared and visible image fusion | |
| Dornaika et al. | A comparative study of image segmentation algorithms and descriptors for building detection | |
| CN111222576A (en) | High-resolution remote sensing image classification method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180403 |