CN117692649B - Ship remote monitoring video efficient transmission method based on image feature matching - Google Patents
Ship remote monitoring video efficient transmission method based on image feature matching Download PDFInfo
- Publication number
- CN117692649B CN117692649B CN202410146672.7A CN202410146672A CN117692649B CN 117692649 B CN117692649 B CN 117692649B CN 202410146672 A CN202410146672 A CN 202410146672A CN 117692649 B CN117692649 B CN 117692649B
- Authority
- CN
- China
- Prior art keywords
- feature
- points
- point
- target
- frame
- 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.)
- Active
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 55
- 230000005540 biological transmission Effects 0.000 title claims abstract description 29
- 230000008859 change Effects 0.000 claims abstract description 74
- 230000006835 compression Effects 0.000 claims abstract description 19
- 238000007906 compression Methods 0.000 claims abstract description 19
- 230000011218 segmentation Effects 0.000 claims description 9
- 238000001514 detection method Methods 0.000 claims description 7
- 230000002596 correlated effect Effects 0.000 claims description 6
- 230000000875 corresponding effect Effects 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000004806 packaging method and process Methods 0.000 claims description 2
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 10
- 230000008569 process Effects 0.000 description 20
- 239000007921 spray Substances 0.000 description 15
- 238000012545 processing Methods 0.000 description 4
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 230000001174 ascending effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 239000011347 resin Substances 0.000 description 1
- 229920005989 resin Polymers 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/167—Position within a video image, e.g. region of interest [ROI]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/136—Incoming video signal characteristics or properties
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the field of monitoring video compression transmission, in particular to a ship remote monitoring video efficient transmission method based on image feature matching. According to the method, firstly, a gray image of each frame of a ship monitoring video is extracted, key points and corresponding feature descriptors in the gray image of each frame are extracted, key points of all gray images are matched, a plurality of matching point sequences and feature points in the gray images are obtained, structural features and local information features of the feature points in a preset neighborhood range are analyzed by combining single-frame gray images and multi-frame gray images, the feature points in each frame of the gray images are clustered through the obtained neighborhood change degree, an interested region of each frame of the gray images is obtained, the interested region and a non-interested region of each frame of the gray images are compressed in different modes, and all compressed images are transmitted. The invention reduces the volume of the ship monitoring video, maintains the precision of key information and improves the transmission effect of the monitoring video.
Description
Technical Field
The invention relates to the field of monitoring video compression transmission, in particular to a ship remote monitoring video efficient transmission method based on image feature matching.
Background
The ship remote monitoring system can monitor sea surface conditions of nearby sea areas in the ship navigation process in real time, transmits monitoring videos to the shore side monitoring and dispatching center, timely finds potential safety hazards of other ships or reefs and the like, and makes a reaction faster and takes corresponding measures, so that the high-efficiency transmission of the ship remote monitoring videos is vital for guaranteeing the ship navigation safety.
The method has the advantages that the communication graph splitting and clustering algorithm is utilized to analyze and process the ship remote monitoring video frame images, morphological characteristics of interesting areas such as ships or reefs in the nearby sea areas can be effectively analyzed to determine compression strategies, so that the transmission of the monitoring video is realized, but due to frequent occurrence of sea surface spoons, the spoons can be mistakenly recognized as the interesting areas in the prior art, the video space is reduced through compression, meanwhile, the precision of key information in the video cannot be guaranteed, and the transmission effect of the ship remote monitoring video is poor.
Disclosure of Invention
In order to solve the technical problem that the prior art can mistakenly identify a spray region as an interested region, the precision of key information in a video can not be ensured while the video space is reduced by compression, so that the transmission effect of a ship remote monitoring video is poor, the invention aims to provide a ship remote monitoring video efficient transmission method based on image feature matching, and the adopted technical scheme is as follows:
the invention provides a high-efficiency transmission method of a ship remote monitoring video based on image feature matching, which comprises the following steps:
Acquiring gray level images of each frame in a video for monitoring sea surface conditions of a ship, and carrying out local detection on each frame of gray level images to obtain key points and feature descriptors of the key points in each frame of gray level images;
Matching key points in all gray images according to the feature descriptors to obtain different matching point sequences and feature points in the gray images; selecting one feature point in any frame of gray level image as a target feature point, and obtaining a neighborhood distribution index of the target feature point according to the distance between the target feature point and other feature points and the difference of the feature descriptors in the preset field range of the target feature point; obtaining the stability of the target feature points according to the difference of the neighborhood distribution indexes between adjacent feature points in the matching point sequence where the target feature points are located;
obtaining the variation similarity of the target feature points according to the difference of the stability between the target feature points and other feature points in a preset neighborhood range; obtaining the neighborhood change degree of the target feature point according to the difference of the change similarity between the target feature point and other feature points in a preset neighborhood range and an included angle formed by a connecting line between the target feature point and the other feature points and a horizontal line; clustering the characteristic points in each frame of gray level image based on the neighborhood change degree to obtain an interested region of each frame of gray level image;
And compressing the interested region and the non-interested region of each frame of gray level image in different modes to obtain compressed images of each frame, and transmitting all the compressed images.
Further, in the preset field range of the target feature point, obtaining the neighborhood distribution index of the target feature point according to the distance between the target feature point and other feature points and the difference of the feature descriptors includes:
In a preset neighborhood range, taking the distance between each other characteristic point and the target characteristic point as a distance parameter of each other characteristic point; sequencing the other feature points according to the sequence from small to large of the distance parameters to obtain a neighborhood feature point sequence;
taking the combination of every two adjacent characteristic points in the neighborhood characteristic point sequence as a first characteristic point group;
Taking the absolute value of the difference value of the distance parameters of the two feature points in each first feature point group as the distance difference of each first feature point group; cosine similarity of feature descriptors of two feature points in each first feature point group is used as feature similarity of each first feature point group;
Taking the product value of the distance difference and the feature similarity as a distribution parameter of each first feature point group;
and taking the average value of the distribution parameters of all the first characteristic point groups as a neighborhood distribution index of the target characteristic points.
Further, the obtaining the stability of the target feature point according to the difference of the neighborhood distribution indexes between the adjacent feature points in the matching point sequence where the target feature point is located includes:
The combination of every two adjacent characteristic points in the matching point sequence where the target characteristic points are located is used as a second characteristic point group;
Taking the absolute value of the difference value of the neighborhood distribution indexes of the two characteristic points in each second characteristic point group as the index difference of each second characteristic point group;
And carrying out negative correlation normalization on the average value of the index differences of all the second characteristic point groups to obtain the stability of the target characteristic points.
Further, obtaining the similarity of the target feature point according to the difference of the stability between the target feature point and other feature points in the preset neighborhood range includes:
Inputting the stability of all the characteristic points in the gray image where the target characteristic points are located into a maximum inter-class variance algorithm for calculation to obtain a stability segmentation threshold;
Taking the difference value between the stability of the target characteristic point and the stability segmentation threshold value as a first coefficient, taking the difference value between the stability of each other characteristic point in a preset neighborhood range and the stability segmentation threshold value as a second coefficient, and taking the ratio of the first coefficient to the second coefficient as a judgment parameter; performing negative correlation mapping on the judging parameters to obtain adjustment parameters between the target feature points and each other feature point;
taking the absolute value of the difference value between the stability of the target feature point and the stability of each other feature point in the preset neighborhood range as the initial difference value between the target feature point and each other feature point;
Taking the product value of the adjustment parameter and the initial difference as the adjustment difference between the target feature point and each other feature point;
carrying out negative correlation normalization on the adjustment difference degree to obtain similarity parameters between the target feature point and each other feature point;
and taking the average value of the similarity parameters between the target feature point and all other feature points as the change similarity of the target feature point.
Further, obtaining the neighborhood change degree of the target feature point according to the difference of the change similarity between the target feature point and other feature points in the preset neighborhood range and an included angle formed by a connecting line between the target feature point and the other feature points and a horizontal line includes:
Taking the absolute value of the difference value between the change similarity of each other characteristic point in the preset neighborhood range and the change similarity of the target characteristic point as a change difference parameter of each other characteristic point in the preset neighborhood range;
The included angle between the connecting line of each other characteristic point in the preset neighborhood range and the target characteristic point and the horizontal line is used as the angle parameter of each other characteristic point in the preset neighborhood range;
taking the combination of any two other characteristic points except the target characteristic point in the preset neighborhood range as a third characteristic point group;
taking the absolute value of the difference value of the variation difference parameters of the two feature points in each third feature point group as the first variation parameter of each third feature point group;
obtaining a second variation parameter of each third characteristic point group by using the absolute value of the difference value of the angle parameters of the two characteristic points in each third characteristic point group;
acquiring initial variation parameters of each third characteristic point group, wherein the initial variation parameters are positively correlated with the first variation parameters, and the initial variation parameters are negatively correlated with the second variation parameters;
And taking the average value of the initial change parameters of all the third characteristic point groups as the neighborhood change degree of the target characteristic points.
Further, the clustering the feature points in each frame of gray level image based on the neighborhood change degree, and obtaining the interested region of each frame of gray level image includes:
Constructing a connected graph for all feature points in each frame of gray level image based on a connected graph splitting clustering algorithm, acquiring a splitting threshold value of each side in the connected graph, and taking an average value of neighborhood change degrees of two feature points at two ends of each side in the connected graph as a splitting weight coefficient of each side;
taking the product value of the splitting weight coefficient and the splitting threshold value as the optimal splitting threshold value of each side;
based on a connected graph splitting and clustering algorithm, clustering all characteristic points of each frame of gray level image according to the connected graph and an optimal splitting threshold value of each side in the connected graph to obtain different clustering clusters;
taking the average value of the variation similarity of all the characteristic points in each cluster as the cluster center of each cluster;
Taking the cluster with the cluster center larger than a preset threshold value as an interested cluster, and taking the area surrounded by the minimum circumscribed rectangles of all the characteristic points in the interested cluster as an interested area of each frame of gray level image.
Further, the compressing the interested area and the non-interested area of each frame gray level image in different ways, and obtaining the compressed image of each frame includes:
And carrying out lossless compression on the interested region in each frame of gray level image, and carrying out lossy compression on the non-interested region to obtain a compressed image of each frame.
Further, the transmitting all compressed images includes:
Packaging the data of all the compressed images of the monitoring video to generate a video code stream;
and transmitting the video code stream through a transmission network.
Further, the locally detecting the gray level image of each frame to obtain the key point and the feature descriptor of the key point in the gray level image of each frame includes:
And carrying out local detection on each frame of gray level image based on a scale invariant feature conversion algorithm to obtain key points in each frame of gray level image and feature descriptors of each key point.
Further, the matching the key points in all the gray images according to the feature descriptors to obtain different matching point sequences and feature points in the gray images includes:
Based on a scale invariant feature transformation algorithm, matching key points in all gray images according to feature descriptors of the key points to obtain different matching point sequences;
in each frame of gray level image, the key points belonging to the matching point sequence are used as the characteristic points in the corresponding gray level image.
The invention has the following beneficial effects:
according to the method, the situation that the prior art can mistakenly identify a spray region as an interested region is considered, the accuracy of key information in a video cannot be guaranteed while the video space is reduced through compression, and the transmission effect on the monitored video is reduced, so that firstly, gray images of each frame of the ship monitored video are obtained, the possibility that other ships or reefs and the like affect navigation in the interested region and the spray of a background region are considered, meanwhile, the shape of the spray is always in dynamic change in the monitored video, therefore, characteristic points and a plurality of matching point sequences in the gray images can be firstly extracted, the accurate interested region can be conveniently extracted later, the situation that other characteristic point distribution structures around target characteristic points are reflected through the obtained neighborhood distribution index is relatively stable, compared with the spray region, in the process of time lapse, the structural change of the characteristic points in the interested region is relatively stable, the possibility that the target characteristic points belong to the interested region can be reflected through stability, the possibility that the characteristic points belong to the interested region in the preset neighborhood region can be reflected through change similarity, the characteristic point of the interest region can be extracted, the characteristic of the difference of the adjacent region can be obtained through the change of the characteristic points in the neighborhood region, the difference of the adjacent region is compressed, and the accuracy of the monitored video is reduced, and the accuracy of the difference of the characteristic distribution is improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a method for efficiently transmitting a remote monitoring video of a ship based on image feature matching according to an embodiment of the present invention.
Detailed Description
In order to further explain the technical means and effects adopted by the invention to achieve the preset aim, the following is a detailed description of specific implementation, structure, characteristics and effects thereof of the ship remote monitoring video efficient transmission method based on image characteristic matching according to the invention, which is provided by the invention, with reference to the accompanying drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The invention provides a specific scheme of a ship remote monitoring video efficient transmission method based on image feature matching, which is specifically described below with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of a method for efficiently transmitting a remote monitoring video of a ship based on image feature matching according to an embodiment of the present invention is shown, where the method includes:
Step S1: and acquiring gray level images of each frame in the video of the ship for monitoring sea surface conditions, and carrying out local detection on each frame of gray level image to obtain key points and feature descriptors of the key points in each frame of gray level image.
The ship remote monitoring system can monitor sea surface conditions of nearby sea areas in the ship navigation process in real time, transmits monitoring videos to a shore side monitoring scheduling center, timely finds potential safety hazards such as other ships or reefs, reacts faster, takes corresponding measures, analyzes and processes ship remote monitoring video frame images by using a connected graph splitting clustering algorithm in the prior art, can effectively analyze morphological characteristics of interesting areas such as ships or reefs in nearby sea areas to determine compression strategies, and accordingly achieves transmission of the monitoring videos, but due to frequent occurrence of sea surface spray, the prior art can mistakenly identify the spray areas as interesting areas, the video space is reduced through compression, meanwhile, the accuracy of key information in the videos cannot be guaranteed, and the transmission effect of the ship remote monitoring videos is poor.
According to the embodiment of the invention, firstly, a complete monitoring video is obtained from remote monitoring equipment of a ship, the monitoring video is imported into professional video software such as Premiere software, a single-frame image extraction function in the software is utilized to process the imported monitoring video, so that continuous images of each frame of the whole monitoring video are extracted, in order to reduce the calculation amount of subsequent image processing and improve the processing speed, in one embodiment of the invention, the images of each frame are subjected to graying processing and are converted into single-channel gray images, and thus the gray images of each frame of the monitoring video are obtained. It should be noted that the graying process is a technical means well known to those skilled in the art, and will not be described herein.
After the gray image of each frame is obtained, because other ships or reefs and the like existing in the gray image affect navigation areas, the areas of interest are key information for judging potential safety hazards existing in the navigation process of the ships, the accuracy of the areas of interest after the subsequent compression of the images needs to be ensured, meanwhile, the areas of interest have obvious characteristics in the gray image, so that the local detection can be firstly carried out on each frame of image, the key points in each frame of gray image and the characteristic descriptors of each key point are obtained, the characteristic descriptors of the key points and the position structures of the key points can be analyzed in the subsequent steps, the areas of interest in the gray image are accurately extracted, and the compression transmission effect of the monitoring video is improved.
Because the Scale-invariant feature transform algorithm (Scale-INVARIANT FEATURE TRANSFORM, SIFT) can detect local features in an image and extract key points in the image and feature descriptors of each key point, in one embodiment of the present invention, the key points in each frame of gray-Scale image and feature descriptors of each key point are obtained by locally detecting each frame of gray-Scale image based on the Scale-invariant feature transform algorithm, wherein the key points are pixels with more prominent features in the gray-Scale image, the feature descriptors are quantization of local information features of the key points, and the feature descriptors are a vector, and then the key points of each frame and continuous multi-frame gray-Scale image can be analyzed based on the feature descriptors, so that a region of interest in the gray-Scale image can be accurately extracted.
Step S2: matching key points in all gray images according to the feature descriptors to obtain different matching point sequences and feature points in the gray images; selecting one feature point in any frame of gray level image as a target feature point, and obtaining a neighborhood distribution index of the target feature point according to the distance between the target feature point and other feature points and the difference of feature descriptors in the preset field range of the target feature point; and obtaining the stability of the target feature points according to the difference of neighborhood distribution indexes between adjacent feature points in the matching point sequence where the target feature points are located.
Because the marine vessel frequently appears in the sea surface in the sailing process, the spray is a relatively obvious characteristic in the gray level image, and therefore, partial key points extracted through the steps possibly belong to the spray region, so that the subsequent region of interest extracted through a connected graph splitting and clustering algorithm is inaccurate, the transmission effect on the vessel monitoring video is reduced, the key points in all images in the monitoring video are related to each other in consideration of the fact that the feature of the form change of the region of interest in the monitoring video is different with the time, the change of the form or structure of the region of interest in the continuous multi-frame gray level image is not obvious with the time, the change of the form or structure of the spray region in the continuous multi-frame gray level image is relatively obvious with the time due to the sea surface movement, and the feature descriptors in the gray level image can reflect the local information features of the key points.
Because the scale-invariant feature transformation algorithm also has the function of matching key points in different images based on feature descriptors, in one embodiment of the invention, the key points in all gray images are matched according to the feature descriptors of the key points based on the scale-invariant feature transformation algorithm to obtain different matching point sequences; in each frame of gray level image, the key points belonging to the matching point sequence are used as the characteristic points in the corresponding gray level image, so that the elements in the matching point sequence are the characteristic points in different gray level images, the characteristic points in the matching point sequence are arranged in time ascending order, namely, the first characteristic point in the matching point sequence is a certain characteristic point in the first frame of gray level image, the second characteristic point is a certain characteristic point in the second frame of gray level image, and the like, and certain characteristic point only exists in one matching point sequence, and the key point matching based on the scale-invariant characteristic conversion algorithm is a technical means of resin channels of the person skilled in the art, and is not repeated herein.
In order to facilitate analysis of the feature points, one feature point can be selected as a target feature point in any one gray image, in the subsequent analysis process of other feature points in the gray image or the feature points in other gray images, the process of analyzing the target feature points is completely the same, a preset neighborhood range is built for the target feature points, so that analysis of the distribution features of the feature points in the preset neighborhood range is facilitated, wherein the preset neighborhood range is set as a circular area taking the target feature point as the center, the number of other feature points except the target feature point in the preset neighborhood range is set as a preset number, the preset number is set as 15, and specific values of the preset number can also be set by an embodiment according to specific implementation scenes, and the method is not limited; the difference between the feature points in the interested region and the feature points in the spray region is mainly that the relative positions of the feature points in the interested region and the change of the feature descriptors are not obvious in gray images of different frames along with the time, and the relative positions of the feature points in the spray region and the change of the feature descriptors are obvious, so that the differences of the distances between the target feature points and other feature points in a preset neighborhood range and the differences of the feature descriptors can be analyzed firstly, the distribution features of the target feature points in the neighborhood can be quantified through the neighborhood distribution index, the neighborhood distribution index of the feature points in a matching point sequence can be combined conveniently, and the possibility that the target feature points belong to the interested region can be analyzed.
Preferably, in an embodiment of the present invention, the method for obtaining a neighborhood distribution index of a target feature point specifically includes:
In a preset neighborhood range, taking the distance between each other characteristic point and the target characteristic point as a distance parameter of each other characteristic point; the distance parameters are sequenced from small to large, other feature points are sequenced to obtain a neighborhood feature point sequence of the target feature point, and other feature points in a preset neighborhood range can be sequenced from large to small according to the sequence of the distance parameters in other embodiments of the invention, and the method is not limited herein; taking the combination of every two adjacent characteristic points in the neighborhood characteristic point sequence as a first characteristic point group; taking the absolute value of the difference value of the distance parameters of the two feature points in each first feature point group as the distance difference of each first feature point group; cosine similarity of feature descriptors of two feature points in each first feature point group is used as feature similarity of each first feature point group; taking the product value of the distance difference and the feature similarity as the distribution parameter of each first feature point group; and taking the average value of the distribution parameters of all the first characteristic point groups as a neighborhood distribution index of the target characteristic points. The expression of the neighborhood distribution index may specifically be, for example:
wherein, A neighborhood distribution index representing the target feature points; /(I)And/>The/>, in the neighborhood feature point sequence, representing the target feature pointDistance parameters of two feature points in the first feature point group; /(I)And/>The/>, in the neighborhood feature point sequence, representing the target feature pointFeature descriptors of two feature points in the first feature point group, wherein the feature descriptors are multidimensional vectors; /(I)Representing the preset number, namely the number of other characteristic points in a preset neighborhood range of the target characteristic point, and/>Representing the number of the first feature point groups; /(I)Representing a cosine similarity function.
In the process of acquiring the neighborhood distribution index of the target feature point, the neighborhood distribution indexThe method is used for quantifying the characteristics of other characteristic point distribution in a preset neighborhood range of target characteristic points, and the structural characteristics of the relative positions of the characteristic points in the preset neighborhood range are obtained by the distance difference/>To reflect, distance difference/>The larger the feature points in the neighborhood range are, the more scattered the feature points are, and the feature of the local information of the feature points in the preset neighborhood range is obtained through feature similarity/>Reflecting the feature similarityThe larger the local feature of the feature points in the neighborhood range is, the more similar the local feature of the feature points is, and the distribution parameter/>, of the first feature point group is obtained by combining the distance difference and the feature similarityFurther, the average value of the distribution parameters of all the first characteristic point groups is used as a neighborhood distribution index/>。
The obtained neighborhood distribution index can reflect the structural characteristics and the characteristics of local information in the preset neighborhood range of the target characteristic point, the neighborhood distribution index can be obtained through the same method for other characteristic points in the gray level image where the target characteristic point is located and other characteristic points in the gray level image, as the change of the morphology or the structure of the interesting region in the continuous multi-frame gray level image is not obvious along with the time, the difference of the neighborhood distribution index among the characteristic points in the matching point sequence where the target characteristic point is located is small, the wave-flower region is generated due to sea surface movement and is always in dynamic change, the change of the morphology or the structure of the wave-flower region in the continuous multi-frame gray level image is obvious along with the time, the difference of the neighborhood distribution index among the characteristic points in the matching point sequence where the target characteristic point is located is large, and therefore the structural characteristics of the preset neighborhood of the target characteristic point and the change degree of the characteristics of the local information of the interesting region can be reflected through the obtained stability, and the possibility that the target pixel point belongs to the interesting region is reflected.
Preferably, in one embodiment of the present invention, the method for obtaining the stability of the target feature point specifically includes:
The combination of every two adjacent characteristic points in the matching point sequence where the target characteristic points are located is used as a second characteristic point group; taking the absolute value of the difference value of the neighborhood distribution indexes of the two characteristic points in each second characteristic point group as the index difference of each second characteristic point group; and carrying out negative correlation normalization on the average value of the index differences of all the second characteristic point groups to obtain the stability of the target characteristic points. The expression of the stability may specifically be, for example:
wherein, Representing the stability of the target feature points; /(I)And/>Represents the/>, in the matching point sequence where the target feature point is locatedNeighborhood distribution indexes of two feature points in the second feature point group; /(I)Representing the number of feature points in a matching point sequence where the target feature points are located, namely the number of all gray level images in the monitoring video, and/>Representing the number of second feature point groups in the matching point sequence where the target feature points are located; /(I)Expressed as natural constant/>An exponential function of the base.
In the process of acquiring the stability of the target feature points, the stabilityReflecting the change degree of the distribution characteristics of the characteristic points in the preset neighborhood range of the target characteristic points along with the time, and reflecting the possibility that the target characteristic points belong to the region of interest, wherein index difference/>The larger the distribution feature in the preset neighborhood range of the target feature point, the more obvious the change of the distribution feature, and the less likely the target feature point belongs to the region of interest, and the more likely the target feature point belongs to the spray region, so the natural constant/>, is usedNormalizing the average value of index differences of all the second characteristic point groups by using the exponential function as the base to obtain the stability/>, of the target characteristic points。
The corresponding stability of other characteristic points in the gray level image where the target characteristic point is located and the characteristic points in other gray level images can be obtained through the same method, and then the difference of the stability between the characteristic points in the preset neighborhood range of the target characteristic point can be further analyzed, so that the accuracy of identifying the region of interest is improved, and the transmission effect of the monitoring video is further improved.
Step S3: obtaining the variation similarity of the target feature points according to the difference of the stability between the target feature points and other feature points in a preset neighborhood range; obtaining the neighborhood change degree of the target feature point according to the difference of the change similarity between the target feature point and other feature points in a preset neighborhood range and an included angle formed by a connecting line between the target feature point and the other feature points and a horizontal line; and clustering the characteristic points in each frame of gray level image based on the neighborhood change degree to obtain the interested region of each frame of gray level image.
The stability of the target feature point and other feature points, which can be obtained through the steps, reflects the variation degree of the distribution features of the preset neighborhood range of the feature points along with the time, and as the feature points on the region of interest and the feature points on the spray region possibly exist in the preset neighborhood range of the target feature point at the same time, in order to improve the accuracy of the connected graph splitting clustering algorithm in extracting the feature points on the region of interest in the follow-up process, the accuracy of the follow-up clustering of the feature points can be further facilitated by analyzing the difference of the stability between the target feature point and other feature points in the preset neighborhood range and reflecting the similarity degree of the local features and the structural features in the dynamic change between the target feature point and the other feature points in the preset neighborhood range through the obtained variation similarity.
Preferably, in one embodiment of the present invention, the method for obtaining the similarity of the target feature points includes:
The feature points in the gray level image mainly belong to two areas, namely an interested area and a wave-formed area, and the stability can be used for primarily distinguishing the area of the feature points, so that the stability of all the feature points in the gray level image where the target feature points are located can be input into a maximum inter-class variance algorithm for calculation, a stability segmentation threshold value is obtained, and the maximum inter-class variance algorithm is a technical means well known to a person skilled in the art and is not described in detail herein; taking the difference value between the stability of the target characteristic point and the stability segmentation threshold value as a first coefficient, taking the difference value between the stability of each other characteristic point in the preset neighborhood range and the stability segmentation threshold value as a second coefficient, and taking the ratio of the first coefficient to the second coefficient as a judgment parameter; performing negative correlation mapping on the judgment parameters to obtain adjustment parameters between the target feature points and each other feature point, and adjusting the subsequent initial difference degree through the adjustment parameters; taking the absolute value of the difference value between the stability of the target feature point and the stability of each other feature point in the preset neighborhood range as the initial difference value between the target feature point and each other feature point; taking the product value of the adjustment parameter and the initial difference as the adjustment difference between the target feature point and each other feature point; carrying out negative correlation normalization on the adjustment difference degree to obtain similarity parameters between the target feature point and each other feature point; and taking the average value of similarity parameters between the target feature point and all other feature points as the change similarity of the target feature point. The expression of the varying similarity may specifically be, for example:
wherein, Representing the variation similarity of the target feature points; /(I)Representing the target feature points and the/>, within a preset neighborhood rangeSimilarity parameters between the other feature points; /(I)Representing the preset number, namely the number of other characteristic points in a preset neighborhood range of the target characteristic point; /(I)Representing the target feature points and the/>, within a preset neighborhood rangeThe adjustment difference degree among other feature points; /(I)Representing the stability of the target feature points; /(I)Representing the first/>, within a preset neighborhood range of the target feature pointStability of the other feature points; /(I)Representing a stability partition threshold; /(I)Expressed as natural constant/>An exponential function of the base.
In the process of obtaining the change similarity of the target feature points, the change similarityReflecting the similarity degree of the local feature change of the target feature point and the local feature change of other feature points in a preset neighborhood range, and changing the similarity/>The larger the target feature point is, the more similar the local feature changes between the target feature point and other feature points in a preset neighborhood range are, and the larger the probability that the target feature point and other feature points in the preset neighborhood range belong to the region of interest is, wherein the first coefficient/>And a second coefficient/>The same sign indicates that the target feature point and some other feature point in the preset neighborhood belong to the same type of region, and the same sign indicates that the target feature point and the other feature point belong to different types of regions, so that the ratio of the first coefficient to the second coefficient is used as a judgment parameter/>If the judgment parameter is positive, it is stated that the target feature point and the other feature points belong to the same type of region, the target feature point is obtained by using a natural constant/>The base exponential function carries out negative correlation mapping on the judgment parameters to obtain adjustment parameters/>Reducing initial difference/>, between target feature points and other feature points by using adjustment parametersTo reduce the adjustment difference between the target feature point and other feature points/>Increasing similarity parameters/>, between target feature points and other feature pointsThe similarity parameter between the feature points belonging to the same type of region is larger, if the judgment parameter is negative, the initial difference/>, between the target feature point and other feature points is increased by using the adjustment parameter when the judgment parameter indicates that the target feature point and the other feature points belong to different types of regionsTo reduce the similarity parameter/>, between the target feature point and other feature pointsFurther, the average value of similarity parameters between the target feature point and all other feature points is taken as the variation similarity/>, of the target feature point。
According to the embodiment of the invention, the feature points are required to be clustered by using a connected graph splitting and clustering algorithm in the subsequent steps, so that the feature points belonging to the region of interest are extracted, the connected graph is firstly constructed for the feature points in the gray image by taking the connected graph splitting and clustering algorithm into consideration, the directional change of the region exists in the feature points belonging to the region of interest on the structural distribution characteristic of the connected graph, the characteristic that the change similarity of the other feature points in the preset neighborhood range of the feature points belonging to the region of interest is directional and aggregated towards the inside of the region of interest is particularly shown, the aggregation distribution characteristic is mainly caused by the difference of the stability of the feature points between the region of interest and the spray region, and therefore, the difference of the change similarity between the target feature points and the other feature points in the preset neighborhood range and the included angle formed by connecting lines and horizontal lines between the target feature points can be analyzed, the accuracy of the subsequent clustering is improved through the obtained neighborhood change degree, and the accuracy of the extraction of the region of interest in the image is further improved.
Preferably, in an embodiment of the present invention, the method for obtaining the neighborhood change degree of the target feature point specifically includes:
Taking the absolute value of the difference value between the change similarity of each other characteristic point in the preset neighborhood range and the change similarity of the target characteristic point as a change difference parameter of each other characteristic point in the preset neighborhood range; the included angle between the connecting line of each other characteristic point in the preset neighborhood range and the target characteristic point and the horizontal line is used as the angle parameter of each other characteristic point in the preset neighborhood range; taking the combination of any two other characteristic points except the target characteristic point in the preset neighborhood range as a third characteristic point group; taking the absolute value of the difference value of the variation difference parameters of the two feature points in each third feature point group as the first variation parameter of each third feature point group; obtaining a second variation parameter of each third characteristic point group by using the absolute value of the difference value of the angle parameters of the two characteristic points in each third characteristic point group; acquiring initial variation parameters of each third characteristic point group, wherein the initial variation parameters are positively correlated with the first variation parameters, and the initial variation parameters are negatively correlated with the second variation parameters; and taking the average value of the initial change parameters of all the third characteristic point groups as the neighborhood change degree of the target characteristic points. The expression of the neighborhood change degree may specifically be, for example:
wherein, Representing the neighborhood change degree of the target feature points; /(I)Representing the first neighbor region of the target feature pointVariation difference parameters of other feature points; /(I)And/>Representing the first/>, within a preset neighborhood range of the target feature pointA variation difference parameter of two feature points in the third feature point group; /(I)And/>Representing the first/>, within a preset neighborhood range of the target feature pointAngle parameters of two characteristic points in the third characteristic point group; /(I)Representing the variation similarity of the target feature points; /(I)Representing the first/>, within a preset neighborhood range of the target feature pointThe varying similarity of the other feature points; /(I)Representing permutation and combination functions, then/>Representing the number of combinations of two feature points arbitrarily selected from all other feature points within a preset neighborhood range of the target feature point, namely/>Representing the number of the third feature point groups; /(I)The adjustment parameter is used for preventing the denominator from being 0, the adjustment parameter is set to be 0.01, and specific numerical values of the adjustment parameter can also be set by an implementer according to specific implementation scenarios, and are not limited herein.
In the process of obtaining the neighborhood change degree of the target feature points, the neighborhood change degreeIn the process of using a connected graph splitting and clustering algorithm subsequently, the splitting threshold value of the edge related to the target feature point in the connected graph is weighted and adjusted, and because the change similarity of other feature points in the preset neighborhood range of the feature point of the region of interest has the characteristic of directional aggregation towards the interior of the region of interest, the first change parameter/>The larger the difference of the variation similarity between the target feature point and the two feature points in the third feature point group is, the larger the neighborhood variation degree/>The larger the second variation parameter/>Representing the included angle formed by the connection line of two characteristic points in the third characteristic point group and the target characteristic point, wherein the smaller the second change parameter is, the smaller the included angle formed by the three characteristic points is, the neighborhood change degree/>The larger and further the initial variation parameters/>, of all third feature point groupsIs taken as the neighborhood change degree/>, of the target feature point。
According to the embodiment of the invention, the feature points in the gray image are clustered by using the connected graph splitting and clustering algorithm to extract the region of interest, but the accuracy of extracting the feature points of the region of interest can be reduced in the process of using the connected graph splitting and clustering algorithm due to the existence of the spoondrift region, and the feature points in the region of interest cannot be effectively distinguished from the feature points in the spoondrift region after clustering due to unreasonable setting of the splitting threshold value of each side in the constructed connected graph in the connected graph splitting and clustering algorithm, so that the acquired neighborhood variation is required to be used for adjusting the splitting threshold value of each side in the connected graph, and the constructed connected graph is accurately split by using the connected graph splitting and clustering algorithm, so that the region of interest in each frame of gray image is extracted.
Preferably, in one embodiment of the present invention, the method for acquiring the region of interest of each frame of gray-scale image specifically includes:
constructing a connected graph for all feature points in each frame of gray level image based on a connected graph split clustering algorithm, and acquiring a split threshold value of each side in the connected graph, wherein the split threshold value of each side in the connected graph can be set through the connected domain split clustering algorithm, and an average value of the neighborhood change degree of two feature points at two ends of each side in the connected graph is used as a split weight coefficient of each side; taking the product value of the splitting weight coefficient and the splitting threshold value as the optimal splitting threshold value of each side; based on a connected graph splitting and clustering algorithm, clustering all characteristic points of each frame of gray level image according to the connected graph and an optimal splitting threshold value of each side in the connected graph to obtain different clustering clusters; the larger the variation similarity of the feature points is, the average value of the variation similarity of all the feature points in each cluster is used as the cluster center of each cluster; the clustering cluster with the clustering center larger than the preset threshold value is used as an interested clustering cluster, the area surrounded by the smallest circumscribed rectangles of all the characteristic points in the interested clustering cluster is used as an interested area of each frame of gray level image, the preset threshold value is set to be 0.6, and the specific value of the preset threshold value can be set by an implementer according to specific implementation scenes and is not limited.
After the interested area of each frame of gray level image is obtained, different modes of compression can be carried out on different areas in each frame of gray level image in the follow-up process, so that the accuracy of key information in a monitoring video is ensured to be higher while the space of the ship monitoring video is reduced.
Step S4: and compressing the interested region and the non-interested region of each frame of gray level image in different modes to obtain compressed images of each frame, and transmitting all the compressed images.
The region of interest in each frame of gray level image is the region of other ships or reefs which affect ship navigation, the region is the key information in the monitoring video, the background region except the region of interest in each frame of gray level image is the non-region of interest, the key of the information of the non-region of interest to the navigation safety of the monitoring ship is lower, so that the precision of key information in the monitoring video is required to be ensured to be higher while the video space is reduced by compressing the monitoring video, the region of interest and the non-region of interest of each frame of gray level image can be compressed in different modes to obtain compressed images of each frame, all compressed images are transmitted, and the transmission efficiency of the monitoring video is improved while the key information in the monitoring video is also ensured not to be damaged.
Preferably, the method for acquiring a compressed image of each frame in one embodiment of the present invention specifically includes:
Because the interested region in the gray level image is the key information, the interested region in each frame of gray level image can be subjected to lossless compression, and the non-interested region can be subjected to lossy compression, so that a compressed image of each frame can be obtained, wherein the lossless compression and the lossy compression are all technical means well known to those skilled in the art, and are not described herein.
After the compressed image of each frame is obtained, the data of all the compressed images of the monitoring video can be packaged, a video code stream is generated, and the video code stream is transmitted through a transmission network.
In summary, the embodiment of the invention firstly obtains the gray level image of each frame in the ship monitoring video, carries out local detection on each frame of gray level image, obtains the key points in each frame of gray level image and the feature descriptors of the key points, matches the key points in all gray level images, and obtains different matching point sequences and feature points in the gray level image; selecting one feature point in any frame of gray level image as a target feature point, and obtaining a neighborhood distribution index of the target feature point according to the distance between the target feature point and other feature points and the difference of feature descriptors in the preset field range of the target feature point; obtaining the stability of the target feature points according to the difference of neighborhood distribution indexes between adjacent feature points in the matching point sequence where the target feature points are located; obtaining the variation similarity of the target feature points according to the difference of the stability between the target feature points and other feature points in a preset neighborhood range; obtaining the neighborhood change degree of the target feature point according to the difference of the change similarity between the target feature point and other feature points in a preset neighborhood range and an included angle formed by a connecting line between the target feature point and the other feature points and a horizontal line; clustering characteristic points in each frame of gray level image based on the neighborhood change degree to obtain an interested region of each frame of gray level image; and compressing the interested region and the non-interested region of each frame of gray level image in different modes to obtain compressed images of each frame, and transmitting all the compressed images.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.
Claims (10)
1. The efficient transmission method of the ship remote monitoring video based on image feature matching is characterized by comprising the following steps of:
Acquiring gray level images of each frame in a video for monitoring sea surface conditions of a ship, and carrying out local detection on each frame of gray level images to obtain key points and feature descriptors of the key points in each frame of gray level images;
Matching key points in all gray images according to the feature descriptors to obtain different matching point sequences and feature points in the gray images; selecting one feature point in any frame of gray level image as a target feature point, and obtaining a neighborhood distribution index of the target feature point according to the distance between the target feature point and other feature points and the difference of the feature descriptors in the preset field range of the target feature point; obtaining the stability of the target feature points according to the difference of the neighborhood distribution indexes between adjacent feature points in the matching point sequence where the target feature points are located;
obtaining the variation similarity of the target feature points according to the difference of the stability between the target feature points and other feature points in a preset neighborhood range; obtaining the neighborhood change degree of the target feature point according to the difference of the change similarity between the target feature point and other feature points in a preset neighborhood range and an included angle formed by a connecting line between the target feature point and the other feature points and a horizontal line; clustering the characteristic points in each frame of gray level image based on the neighborhood change degree to obtain an interested region of each frame of gray level image;
And compressing the interested region and the non-interested region of each frame of gray level image in different modes to obtain compressed images of each frame, and transmitting all the compressed images.
2. The method for efficiently transmitting the remote monitoring video of the ship based on the image feature matching according to claim 1, wherein the obtaining the neighborhood distribution index of the target feature point according to the distance between the target feature point and other feature points and the difference of the feature descriptors in the preset field range of the target feature point comprises:
In a preset neighborhood range, taking the distance between each other characteristic point and the target characteristic point as a distance parameter of each other characteristic point; sequencing the other feature points according to the sequence from small to large of the distance parameters to obtain a neighborhood feature point sequence;
taking the combination of every two adjacent characteristic points in the neighborhood characteristic point sequence as a first characteristic point group;
Taking the absolute value of the difference value of the distance parameters of the two feature points in each first feature point group as the distance difference of each first feature point group; cosine similarity of feature descriptors of two feature points in each first feature point group is used as feature similarity of each first feature point group;
Taking the product value of the distance difference and the feature similarity as a distribution parameter of each first feature point group;
and taking the average value of the distribution parameters of all the first characteristic point groups as a neighborhood distribution index of the target characteristic points.
3. The method for efficiently transmitting the remote monitoring video of the ship based on the image feature matching according to claim 1, wherein the obtaining the stability of the target feature point according to the difference of the neighborhood distribution indexes between the adjacent feature points in the matching point sequence where the target feature point is located comprises:
The combination of every two adjacent characteristic points in the matching point sequence where the target characteristic points are located is used as a second characteristic point group;
Taking the absolute value of the difference value of the neighborhood distribution indexes of the two characteristic points in each second characteristic point group as the index difference of each second characteristic point group;
And carrying out negative correlation normalization on the average value of the index differences of all the second characteristic point groups to obtain the stability of the target characteristic points.
4. The method for efficiently transmitting the remote monitoring video of the ship based on the image feature matching according to claim 1, wherein the obtaining the variation similarity of the target feature point according to the difference of the stability between the target feature point and other feature points in a preset neighborhood range comprises:
Inputting the stability of all the characteristic points in the gray image where the target characteristic points are located into a maximum inter-class variance algorithm for calculation to obtain a stability segmentation threshold;
Taking the difference value between the stability of the target characteristic point and the stability segmentation threshold value as a first coefficient, taking the difference value between the stability of each other characteristic point in a preset neighborhood range and the stability segmentation threshold value as a second coefficient, and taking the ratio of the first coefficient to the second coefficient as a judgment parameter; performing negative correlation mapping on the judging parameters to obtain adjustment parameters between the target feature points and each other feature point;
taking the absolute value of the difference value between the stability of the target feature point and the stability of each other feature point in the preset neighborhood range as the initial difference value between the target feature point and each other feature point;
Taking the product value of the adjustment parameter and the initial difference as the adjustment difference between the target feature point and each other feature point;
carrying out negative correlation normalization on the adjustment difference degree to obtain similarity parameters between the target feature point and each other feature point;
and taking the average value of the similarity parameters between the target feature point and all other feature points as the change similarity of the target feature point.
5. The method for efficiently transmitting the remote monitoring video of the ship based on the image feature matching according to the claim 1, wherein the obtaining the neighborhood change degree of the target feature point according to the difference of the change similarity between the target feature point and other feature points in the preset neighborhood range and the included angle formed by the connecting line between the target feature point and the other feature points and the horizontal line comprises:
Taking the absolute value of the difference value between the change similarity of each other characteristic point in the preset neighborhood range and the change similarity of the target characteristic point as a change difference parameter of each other characteristic point in the preset neighborhood range;
The included angle between the connecting line of each other characteristic point in the preset neighborhood range and the target characteristic point and the horizontal line is used as the angle parameter of each other characteristic point in the preset neighborhood range;
taking the combination of any two other characteristic points except the target characteristic point in the preset neighborhood range as a third characteristic point group;
taking the absolute value of the difference value of the variation difference parameters of the two feature points in each third feature point group as the first variation parameter of each third feature point group;
taking the absolute value of the difference value of the angle parameters of the two feature points in each third feature point group as a second variation parameter of each third feature point group;
acquiring initial variation parameters of each third characteristic point group, wherein the initial variation parameters are positively correlated with the first variation parameters, and the initial variation parameters are negatively correlated with the second variation parameters;
And taking the average value of the initial change parameters of all the third characteristic point groups as the neighborhood change degree of the target characteristic points.
6. The efficient transmission method of the ship remote monitoring video based on image feature matching according to claim 1, wherein the clustering the feature points in each frame of gray level image based on the neighborhood change degree, to obtain the region of interest of each frame of gray level image comprises:
Constructing a connected graph for all feature points in each frame of gray level image based on a connected graph splitting clustering algorithm, acquiring a splitting threshold value of each side in the connected graph, and taking an average value of neighborhood change degrees of two feature points at two ends of each side in the connected graph as a splitting weight coefficient of each side;
taking the product value of the splitting weight coefficient and the splitting threshold value as the optimal splitting threshold value of each side;
based on a connected graph splitting and clustering algorithm, clustering all characteristic points of each frame of gray level image according to the connected graph and an optimal splitting threshold value of each side in the connected graph to obtain different clustering clusters;
taking the average value of the variation similarity of all the characteristic points in each cluster as the cluster center of each cluster;
Taking the cluster with the cluster center larger than a preset threshold value as an interested cluster, and taking the area surrounded by the minimum circumscribed rectangles of all the characteristic points in the interested cluster as an interested area of each frame of gray level image.
7. The method for efficiently transmitting the ship remote monitoring video based on image feature matching according to claim 1, wherein the compressing the interested region and the non-interested region of each frame of gray level image in different ways to obtain the compressed image of each frame comprises:
And carrying out lossless compression on the interested region in each frame of gray level image, and carrying out lossy compression on the non-interested region to obtain a compressed image of each frame.
8. The efficient transmission method of the ship remote monitoring video based on image feature matching according to claim 1, wherein the transmitting all compressed images comprises:
Packaging the data of all the compressed images of the monitoring video to generate a video code stream;
and transmitting the video code stream through a transmission network.
9. The method for efficiently transmitting the ship remote monitoring video based on image feature matching according to claim 1, wherein the step of locally detecting each frame of gray level image to obtain the key points and the feature descriptors of the key points in each frame of gray level image comprises the following steps:
And carrying out local detection on each frame of gray level image based on a scale invariant feature conversion algorithm to obtain key points in each frame of gray level image and feature descriptors of each key point.
10. The efficient transmission method of the ship remote monitoring video based on image feature matching according to claim 1, wherein the matching of key points in all gray images according to the feature descriptors to obtain different matching point sequences and feature points in gray images comprises:
Based on a scale invariant feature transformation algorithm, matching key points in all gray images according to feature descriptors of the key points to obtain different matching point sequences;
in each frame of gray level image, the key points belonging to the matching point sequence are used as the characteristic points in the corresponding gray level image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410146672.7A CN117692649B (en) | 2024-02-02 | 2024-02-02 | Ship remote monitoring video efficient transmission method based on image feature matching |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410146672.7A CN117692649B (en) | 2024-02-02 | 2024-02-02 | Ship remote monitoring video efficient transmission method based on image feature matching |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117692649A CN117692649A (en) | 2024-03-12 |
| CN117692649B true CN117692649B (en) | 2024-04-19 |
Family
ID=90132330
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410146672.7A Active CN117692649B (en) | 2024-02-02 | 2024-02-02 | Ship remote monitoring video efficient transmission method based on image feature matching |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117692649B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117935125B (en) * | 2024-03-20 | 2024-05-28 | 常熟理工学院 | Intelligent elevator behavior recognition method based on artificial intelligence |
| CN117934571B (en) * | 2024-03-21 | 2024-06-07 | 广州市艾索技术有限公司 | 4K high-definition KVM seat management system |
| CN118014887B (en) * | 2024-04-10 | 2024-06-04 | 佰鸟纵横科技(天津)有限公司 | Intelligent folding house remote monitoring method based on Internet of things |
| CN118368433B (en) * | 2024-06-20 | 2024-09-17 | 深圳金三立视频科技股份有限公司 | Video coding compression method based on bandwidth self-adaption |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN205610843U (en) * | 2016-02-03 | 2016-09-28 | 广州中海电信有限公司 | Ocean vessel satellite communication video data compression box and system |
| WO2017181892A1 (en) * | 2016-04-19 | 2017-10-26 | 中兴通讯股份有限公司 | Foreground segmentation method and device |
| CN110097093A (en) * | 2019-04-15 | 2019-08-06 | 河海大学 | A kind of heterologous accurate matching of image method |
| CN113191266A (en) * | 2021-04-30 | 2021-07-30 | 江苏航运职业技术学院 | Remote monitoring management method and system for ship power device |
| CN114973028A (en) * | 2022-05-17 | 2022-08-30 | 中国电子科技集团公司第十研究所 | Aerial video image real-time change detection method and system |
| WO2023025236A1 (en) * | 2021-08-26 | 2023-03-02 | 交通运输部水运科学研究所 | Multi-navigation-element data fusion method |
| CN115797607A (en) * | 2023-02-13 | 2023-03-14 | 无锡文康科技有限公司 | Image optimization processing method for enhancing VR real effect |
| CN115984284A (en) * | 2023-03-21 | 2023-04-18 | 黄海造船有限公司 | X-ray hull weld image detection method for ship maintenance |
| CN116340801A (en) * | 2023-05-31 | 2023-06-27 | 烟台市福山区动物疫病预防控制中心 | Intelligent monitoring method for abnormal environmental data of livestock breeding |
| CN117372966A (en) * | 2023-12-06 | 2024-01-09 | 陕西龙跃锐星科技有限公司 | Turntable state monitoring method based on video monitoring |
| CN117456428A (en) * | 2023-12-22 | 2024-01-26 | 杭州臻善信息技术有限公司 | Garbage throwing behavior detection method based on video image feature analysis |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5891409B2 (en) * | 2012-01-12 | 2016-03-23 | パナソニックＩｐマネジメント株式会社 | Feature extraction device, feature extraction method, and feature extraction program |
| EP3777152A4 (en) * | 2019-06-04 | 2021-02-17 | SZ DJI Technology Co., Ltd. | Method, device, and storage medium for encoding video data base on regions of interests |
-
2024
- 2024-02-02 CN CN202410146672.7A patent/CN117692649B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN205610843U (en) * | 2016-02-03 | 2016-09-28 | 广州中海电信有限公司 | Ocean vessel satellite communication video data compression box and system |
| WO2017181892A1 (en) * | 2016-04-19 | 2017-10-26 | 中兴通讯股份有限公司 | Foreground segmentation method and device |
| CN110097093A (en) * | 2019-04-15 | 2019-08-06 | 河海大学 | A kind of heterologous accurate matching of image method |
| CN113191266A (en) * | 2021-04-30 | 2021-07-30 | 江苏航运职业技术学院 | Remote monitoring management method and system for ship power device |
| WO2023025236A1 (en) * | 2021-08-26 | 2023-03-02 | 交通运输部水运科学研究所 | Multi-navigation-element data fusion method |
| CN114973028A (en) * | 2022-05-17 | 2022-08-30 | 中国电子科技集团公司第十研究所 | Aerial video image real-time change detection method and system |
| CN115797607A (en) * | 2023-02-13 | 2023-03-14 | 无锡文康科技有限公司 | Image optimization processing method for enhancing VR real effect |
| CN115984284A (en) * | 2023-03-21 | 2023-04-18 | 黄海造船有限公司 | X-ray hull weld image detection method for ship maintenance |
| CN116340801A (en) * | 2023-05-31 | 2023-06-27 | 烟台市福山区动物疫病预防控制中心 | Intelligent monitoring method for abnormal environmental data of livestock breeding |
| CN117372966A (en) * | 2023-12-06 | 2024-01-09 | 陕西龙跃锐星科技有限公司 | Turntable state monitoring method based on video monitoring |
| CN117456428A (en) * | 2023-12-22 | 2024-01-26 | 杭州臻善信息技术有限公司 | Garbage throwing behavior detection method based on video image feature analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117692649A (en) | 2024-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117692649B (en) | Ship remote monitoring video efficient transmission method based on image feature matching | |
| CN109871902B (en) | SAR small sample identification method based on super-resolution countermeasure generation cascade network | |
| CN114022793B (en) | Optical remote sensing image change detection method based on twin network | |
| CN105528794A (en) | Moving object detection method based on Gaussian mixture model and superpixel segmentation | |
| CN115914649A (en) | Data transmission method and system for medical video | |
| CN116092013B (en) | Dangerous road condition identification method for intelligent monitoring | |
| CN111915015B (en) | Abnormal value detection method and device, terminal equipment and storage medium | |
| CN107992818B (en) | Method for detecting sea surface ship target by optical remote sensing image | |
| CN115527269B (en) | Intelligent human body posture image recognition method and system | |
| CN111950563A (en) | Image matching method and device and computer readable storage medium | |
| CN116233479A (en) | Live broadcast information content auditing system and method based on data processing | |
| CN115456113A (en) | Modulation format identification method based on constellation diagram multi-feature extraction algorithm | |
| CN109947960B (en) | Face multi-attribute joint estimation model construction method based on depth convolution | |
| Luqman et al. | Subgraph spotting through explicit graph embedding: An application to content spotting in graphic document images | |
| US20110150344A1 (en) | Content based image retrieval apparatus and method | |
| CN112699901A (en) | Plant species identification system based on Internet of things | |
| CN110766655A (en) | Hyperspectral image significance analysis method based on abundance | |
| CN114119577B (en) | High-speed rail tunnel leakage cable buckle detection method | |
| CN109241886B (en) | Face recognition method and system based on OLBP and PCA | |
| Han et al. | Comparison of distance measures on Fuzzyc-means algorithm for image classification problem | |
| CN111625672B (en) | Image processing method, image processing device, computer equipment and storage medium | |
| CN117853937B (en) | Rice disease identification method and system based on secondary color cluster analysis | |
| CN117671181B (en) | Method and system for constructing smart city contracture map based on big data | |
| Huang et al. | Research on Optimization Retrieval Technology of Multimedia Information Management System Based on Fuzzy Control | |
| CN114332498A (en) | Multi-size image change detection device and method based on multi-feature extraction |
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 |