CN116109682A - Image registration method based on image diffusion characteristics - Google Patents
Image registration method based on image diffusion characteristics Download PDFInfo
- Publication number
- CN116109682A CN116109682A CN202211163413.2A CN202211163413A CN116109682A CN 116109682 A CN116109682 A CN 116109682A CN 202211163413 A CN202211163413 A CN 202211163413A CN 116109682 A CN116109682 A CN 116109682A
- Authority
- CN
- China
- Prior art keywords
- image
- diffusion
- fft
- dimensional
- fourier spectrum
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000009792 diffusion process Methods 0.000 title claims abstract description 54
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000001228 spectrum Methods 0.000 claims abstract description 29
- 238000012545 processing Methods 0.000 claims abstract description 12
- 238000001514 detection method Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration using non-spatial domain filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20056—Discrete and fast Fourier transform, [DFT, FFT]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to the technical field of image processing, and relates to an image registration method based on image diffusion characteristics, which comprises the following steps: step one: collecting a real-time ground two-dimensional image, and recording the collected ground two-dimensional image as I (x, y), wherein (x, y) is the coordinates of a pixel point in the image; step two: obtaining a two-dimensional Fourier spectrum I of the ground two-dimensional image I (x, y) fft (x, y); step three: diffusion theory based two-dimensional Fourier spectrum I fft (x, y) performing diffusion treatment to obtain a diffused two-dimensional Fourier spectrum I fft_new (x, y); step four: according to the two-dimensional Fourier spectrum I after diffusion fft_new (x, y) obtaining a diffused normal image and outputting the diffused normal image; step five: detecting characteristic points of the diffused normal image and generating a characteristic point description image; step six: describing images with feature pointsAnd matching the pre-stored geographic reference images, and converting the positioning result of the matching result and outputting the positioning result.
Description
Technical Field
The invention relates to a vision enhancement technology and an image matching technology in the field of image processing, in particular to a scene matching positioning method for an unmanned aerial vehicle, and particularly relates to an image registration method based on image diffusion characteristics.
Background
The navigation technology is of great importance in the unmanned system, and is the basis for ensuring whether the unmanned system can execute tasks and return to the navigation safely. In recent years, with the development of computer technology, image processing technology, sensor technology and artificial intelligence technology, as an autonomous, passive, electronic interference resistant, high-resolution visual navigation mode, scene matching visual navigation is increasingly widely applied in unmanned systems.
At present, the conventional method for extracting image features is mostly adopted for solving the problem of rough positioning of top view matching of unmanned aerial vehicles at home and abroad. Some researchers have solved the problem of template matching in aerial images using scale-invariant feature descriptor based techniques. For example, some researchers have investigated the problem of UAV positioning in a GPS-free environment and have utilized optical flow to determine the position of the drone. They use inter-frame transformations to perform gesture tracking and histogram of directional gradient features to register on Google maps and then use particle filtering for finer localization. A more sophisticated solution is to use SIFT features extracted from aerial images as a solution to the UAV image matching problem. In addition, some technical ideas are that an end-to-end image matching method is adopted to complete the positioning of the unmanned aerial vehicle, however, the method depends on a large amount of annotation data, has low interpretation and has high calculation force requirements.
Disclosure of Invention
The invention aims to: the invention provides an image registration method based on image diffusion characteristics for unmanned aerial vehicle visual positioning, which comprises two steps of real-time image preprocessing and matching; the unmanned aerial vehicle system obtains overlook image information of the ground in real time through the photoelectric pod, then converts the image into a frequency domain, calculates diffusion characteristics according to weak differential solution of a diffusion equation, enhances the image information by using the diffusion characteristics, and then outputs an enhancement result by using inverse transformation of fast Fourier transform, so that the preprocessed image is output; the invention has the advantages of low cost, simple operation, low calculation complexity and good image registration effect.
The technical scheme of the invention is as follows:
an image registration method based on image diffusion characteristics comprises the following steps:
step one: collecting a real-time ground two-dimensional image, and recording the collected ground two-dimensional image as I (x, y), wherein (x, y) is the coordinates of a pixel point in the image;
step two: obtaining a two-dimensional Fourier spectrum I of the ground two-dimensional image I (x, y) fft (x,y)；
Step three: diffusion theory based two-dimensional Fourier spectrum I fft (x, y) performing diffusion treatment to obtain a diffused two-dimensional Fourier spectrum I fft_new (x,y)；
Step four: according to the two-dimensional Fourier spectrum I after diffusion fft_new (x, y) obtaining a diffused normal image and outputting the diffused normal image;
step five: detecting characteristic points of the diffused normal image and generating a characteristic point description image;
step six: and matching the characteristic point description image with a pre-stored geographic reference image, and converting the matching result into a positioning result and outputting the positioning result.
Further, in the first step, a real-time ground two-dimensional image is acquired by using a special photoelectric pod device of the unmanned aerial vehicle.
Further, in the third step, the diffusion treatment includes: and calculating the mapping relation between the Diffusion characteristic and the formed image characteristic in the image forming process in a Sobolev space, so as to perform image Diffusion processing.
Further, in the third step, the diffusion process includes the steps of:
step a): calculating a two-dimensional Fourier spectrum I fft The parameter values for extracting the feature at each coordinate (x, y) in (x, y) are as follows:
wherein x is max Is the image height; y is max Is the image width; alpha, beta, lambda and k are all algorithm parameters and are positive numbers;
step b) superimposing the parameter values for extracting features at each coordinate (x, y) into the original two-dimensional Fourier spectrum I fft In (x, y), the formula is as follows:
I fft_new (x,y)＝I fft (x,y)*[T(x,y)+1]
wherein I is fft_new (x, y) is a two-dimensional Fourier spectrum after diffusion processing.
Further, in said step a), the parameter value T (0, 0) =γ at the direct current component (0, 0);
wherein, gamma is an algorithm parameter and is a positive number.
Further, in the second step, a two-dimensional discrete fourier transform is performed on the ground two-dimensional image I (x, y) to obtain a two-dimensional fourier spectrum I thereof fft (x,y)。
Further, in the fourth step, the two-dimensional fourier spectrum I after diffusion fft_new (x, y) performing two-dimensional inverse discrete Fourier transform to obtain a diffused normal image.
In the fifth step, feature point detection is performed on the diffused normal image by using a SIFT algorithm.
The invention has the following beneficial effects:
(1) The image registration method based on the image diffusion characteristics widens the problem solving thought of the similar field, is the first realization of the theory in the field, and has higher research value;
(2) The image registration method based on the image diffusion features, provided by the invention, focuses on signal enhancement of valuable features, can provide a more robust solution for visual positioning in severe weather, and has a larger research potential.
Drawings
FIG. 1 is a schematic flow chart of an image registration method based on image diffusion characteristics;
FIG. 2 is a schematic diagram of a matching operation between a feature point description image obtained by the invention and a pre-stored geographic reference image;
fig. 3 is a schematic view illustrating diffusion processing of an RGB three-channel image according to a first embodiment of the present invention.
Detailed Description
The present invention will be described in detail with reference to specific embodiments. The following examples will assist those skilled in the art in further understanding the present invention, but are not intended to limit the invention in any way. It should be noted that variations and modifications could be made by those skilled in the art without departing from the inventive concept. These are all within the scope of the present invention.
According to the invention, a ground overlook image obtained by the unmanned aerial vehicle in the flight process is taken as an input object, the stable physical quantity which carries the characteristic information in the image is calculated, and how to enhance the semantic information of the image by using the physical quantity with the characteristic information is explored, so that a better image matching effect is achieved, wherein the adopted image enhancement algorithm is mainly realized based on the formation mechanism of image diffusion.
Aiming at the problem that the unmanned aerial vehicle uses image matching to position under the conditions of GPS interruption, interference and the like, the invention designs a novel image enhancement algorithm based on an image diffusion formation mechanism so as to strive for better overlook image matching performance in a general image data set and practical application, and finally achieves the aim that the unmanned aerial vehicle can better realize self-positioning in a real environment only through the acquired overlook image.
The image registration method based on the image diffusion feature of the present invention will be described in detail with reference to fig. 1, and as shown in fig. 1, it includes the following steps:
s11: acquiring a real-time ground overlooking image by using a special photoelectric pod device of the unmanned aerial vehicle, and then reading the ground overlooking image, namely reading an original image acquired by a camera of the photoelectric pod device into a system, wherein the acquired ground two-dimensional image is denoted as I (x, y), and the (x, y) is the coordinates of pixel points in the image;
s12: the read-in image is subjected to a two-dimensional discrete Fourier transform mode to obtain a two-dimensional Fourier spectrum, and the two-dimensional Fourier spectrum is marked as I fft (x,y)；
S13: for the two-dimensional Fourier spectrum I obtained in S12 fft (x, y) performing certain algorithm processing by using a diffusion theory;
in the step, the core technology adopted for processing the two-dimensional Fourier spectrum by using the diffusion theory is as follows: calculating the mapping relation between the Diffuse characteristics and the formed image characteristics in the image forming process in a Sobolev space, thereby deducing a new image processing algorithm based on Diffuse, namely, attributing the problem to an inner product space, and solving the problem in a frequency domain, wherein the specific algorithm is as follows:
1) The parameter values for extracting the features at each coordinate (x, y) are calculated as follows:
wherein x is max Is the image height; y is max Is the image width; alpha, beta, lambda and k are all algorithm parameters and are all positive numbers.
In addition, let the parameter values at the dc component be:
T(0,0)＝γ
wherein γ is also an algorithm parameter.
2) And finally, superposing the calculated parameter values for extracting the characteristics into an original two-dimensional Fourier spectrum, wherein the specific formula is as follows:
I fft_new (x,y)＝I fft (x,y)*[T(x,y)+1]
wherein I is fft_new (x, y) is the processed two-dimensional Fourier spectrum.
S14: for the processed two-dimensional Fourier spectrum I fft_new (x, y) obtaining a normal image by means of two-dimensional inverse discrete fourier transform;
s15: outputting an image processed by a diffusion algorithm; the image processed by the diffusion algorithm can be obtained;
s16: performing feature point detection on the image obtained in the step S15 by using a SIFT algorithm and generating a feature point description image; compared with the conventional feature point detection, after the acquired two-dimensional image is subjected to the diffusion treatment, the feature points in the image are more prominent, so that the feature point description image after the detection by the SIFT algorithm is more explicit, the accuracy of subsequent matching is improved, and the efficiency is improved.
S17: matching the image information obtained in the step S16 with a pre-stored geographic reference image;
s18: and converting and outputting the positioning result of the matching result.
Example 1
With reference to fig. 2 and 3, the image registration is performed on the ground image actually acquired by one unmanned aerial vehicle in the global map image after the ground image is processed by a diffusion equation. Firstly, for diffusion characteristic enhancement operation, an original image is read by adopting RGB color space and split into an R channel, a G channel and a B channel, then two-dimensional images of each channel are subjected to discrete two-dimensional Fourier transform firstly, then corresponding operation is carried out in a frequency domain image, finally the two-dimensional images are processed in a discrete two-dimensional Fourier inverse transform mode, and finally the two-dimensional images after the diffusion characteristic enhancement pretreatment are obtained by combining the two-dimensional images after the diffusion characteristic enhancement pretreatment; and then, for image matching operation, performing feature point detection on the preprocessed image by using a SIFT algorithm in the image processing field, generating corresponding feature point description, and finally matching the feature point description with a pre-stored geographic reference image to realize unmanned aerial vehicle positioning.
The embodiments disclosed herein were chosen and described in detail in order to best explain the principles of the invention and the practical application, and to thereby not limit the invention. Any modifications or variations within the scope of the description that would be apparent to a person skilled in the art are intended to be included within the scope of the invention.
Claims (8)
1. An image registration method based on image diffusion characteristics is characterized in that: the method comprises the following steps:
step one: collecting a real-time ground two-dimensional image, and recording the collected ground two-dimensional image as I (x, y), wherein (x, y) is the coordinates of a pixel point in the image;
step two: obtaining a two-dimensional Fourier spectrum I of the ground two-dimensional image I (x, y) fft (x,y)；
Step three: diffusion theory based two-dimensional Fourier spectrum I fft (x, y) performing diffusion treatment to obtain a diffused two-dimensional Fourier spectrum I fft_new (x,y)；
Step four: according to the two-dimensional Fourier spectrum I after diffusion fft_new (x, y) obtaining a diffused normal image and outputting the diffused normal image;
step five: detecting characteristic points of the diffused normal image and generating a characteristic point description image;
step six: and matching the characteristic point description image with a pre-stored geographic reference image, and converting the matching result into a positioning result and outputting the positioning result.
2. The image registration method based on image diffusion features according to claim 1, wherein: in the first step, a special photoelectric pod device of the unmanned aerial vehicle is used for acquiring a real-time ground two-dimensional image.
3. The image registration method based on image diffusion features according to claim 1, wherein: in the third step, the diffusion treatment includes: and calculating the mapping relation between the Diffusion characteristic and the formed image characteristic in the image forming process in a Sobolev space, so as to perform image Diffusion processing.
4. A method of image registration based on image diffusion features as claimed in claim 3, wherein: in the third step, the diffusion treatment includes the steps of:
step a): calculating a two-dimensional Fourier spectrum I fft The parameter values for extracting the feature at each coordinate (x, y) in (x, y) are as follows:
wherein x is max Is the image height; y is max Is the image width; alpha, beta, lambda and k are all algorithm parameters and are positive numbers;
step b) superimposing the parameter values for extracting features at each coordinate (x, y) into the original two-dimensional Fourier spectrum I fft In (x, y), the formula is as follows:
I fft_new (x,y)＝I fft (x,y)*[T(x,y)+1]
wherein I is fft_new (x, y) is a two-dimensional Fourier spectrum after diffusion processing.
5. The image registration method based on image diffusion features according to claim 4, wherein: in said step a), the parameter value T (0, 0) =γ at the direct current component (0, 0);
wherein, gamma is an algorithm parameter and is a positive number.
6. The image registration method based on image diffusion features according to claim 1, wherein: in the second step, two-dimensional discrete Fourier transform is carried out on the ground two-dimensional image I (x, y) to obtain a two-dimensional Fourier spectrum I thereof fft (x,y)。
7. The image registration method based on image diffusion features according to claim 1, wherein: in the fourth step, the two-dimensional Fourier spectrum I after diffusion is processed fft_new (x, y) performing two-dimensional inverse discrete Fourier transform to obtain a diffused normal image.
8. The image registration method based on image diffusion features according to claim 1, wherein: in the fifth step, feature point detection is performed on the diffused normal image by using a SIFT algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211163413.2A CN116109682A (en) | 2022-09-23 | 2022-09-23 | Image registration method based on image diffusion characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211163413.2A CN116109682A (en) | 2022-09-23 | 2022-09-23 | Image registration method based on image diffusion characteristics |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116109682A true CN116109682A (en) | 2023-05-12 |
Family
ID=86266257
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211163413.2A Pending CN116109682A (en) | 2022-09-23 | 2022-09-23 | Image registration method based on image diffusion characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116109682A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117173222A (en) * | 2023-09-01 | 2023-12-05 | 深圳扬奇医芯智能科技有限公司 | Registration method and system based on diffusion model |
-
2022
- 2022-09-23 CN CN202211163413.2A patent/CN116109682A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117173222A (en) * | 2023-09-01 | 2023-12-05 | 深圳扬奇医芯智能科技有限公司 | Registration method and system based on diffusion model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20220044375A1 (en) | Saliency Map Enhancement-Based Infrared and Visible Light Fusion Method | |
| CN102496157B (en) | Image detection method based on Gaussian multi-scale transform and color complexity | |
| CN102859535B (en) | Daisy descriptor is produced from precalculated metric space | |
| CN109767454B (en) | Unmanned aerial vehicle aerial video moving target detection method based on time-space-frequency significance | |
| CN103632153B (en) | Region-based image saliency map extracting method | |
| CN111414954B (en) | Rock image retrieval method and system | |
| CN117079117B (en) | Underwater image processing and target identification method and device, storage medium and electronic equipment | |
| WO2024077935A1 (en) | Visual-slam-based vehicle positioning method and apparatus | |
| CN112926548A (en) | Lane line detection method and device, electronic equipment and storage medium | |
| CN104050674B (en) | Salient region detection method and device | |
| CN116109682A (en) | Image registration method based on image diffusion characteristics | |
| CN110135435B (en) | Saliency detection method and device based on breadth learning system | |
| Sun et al. | IRDCLNet: Instance segmentation of ship images based on interference reduction and dynamic contour learning in foggy scenes | |
| CN112115778B (en) | Intelligent lane line identification method under ring simulation condition | |
| CN116563553B (en) | Unmanned aerial vehicle image segmentation method and system based on deep learning | |
| Cai et al. | Improving CNN-based planar object detection with geometric prior knowledge | |
| Huang et al. | A Stepwise Refining Image-Level Weakly Supervised Semantic Segmentation Method for Detecting Exposed Surface for Buildings (ESB) From Very High-Resolution Remote Sensing Images | |
| CN108596139B (en) | Remote sensing image urban area extraction method based on Gabor feature saliency | |
| CN110232378B (en) | Image interest point detection method and system and readable storage medium | |
| CN108256519B (en) | Infrared image significance detection method based on global and local interaction | |
| CN111401203A (en) | Target identification method based on multi-dimensional image fusion | |
| CN112348823A (en) | Object-oriented high-resolution remote sensing image segmentation algorithm | |
| Leipnitz et al. | The effect of image resolution in the human presence detection: A case study on real-world image data | |
| Kim et al. | Saliency detection: A self-ordinal resemblance approach | |
| CN110222599A (en) | A kind of gait recognition method based on Gauss Map |
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 |