Search

CN-122023116-A - Unmanned aerial vehicle image-oriented high-fidelity stitching method and system

CN122023116ACN 122023116 ACN122023116 ACN 122023116ACN-122023116-A

Abstract

The invention discloses a high-fidelity splicing method and a system for unmanned aerial vehicle images, wherein a double-stage processing strategy is adopted, accurate positioning is realized through coarse-fine two-stage alignment, coarse alignment is finished through global estimation and calculation of integral transformation parameters, non-rigid offset is refined through local correction and refinement, characteristic weight and geometric smoothness constraint area displacement are dynamically adjusted through combination of self-adaption enhancement, seamless splicing is realized through pixel-level weight masking and multi-band fusion, smooth weights with continuous boundaries and weight sum of 1 are generated based on the shape of an overlapped area, images are decomposed and fused layer by layer in different frequency layers, and the system comprises an image acquisition unit, a double-stage processing unit and an output display unit, and can be deployed at the end side or the ground station of an unmanned aerial vehicle. The invention has high alignment precision, can process large parallax and non-rigid deformation, is integrated with no trace and has a fidelity structure, thereby meeting the real-time processing requirement of airborne or edge equipment.

Inventors

  • LI WANQING
  • SHI XUANMING

Assignees

  • 杭州电子科技大学

Dates

Publication Date
20260512
Application Date
20251230

Claims (7)

  1. 1. The high-fidelity splicing method for the unmanned aerial vehicle image is characterized by comprising the following steps of: The method comprises the steps of firstly, in an alignment stage, realizing coarse alignment by calculating integral transformation parameters between two images, and further estimating local non-rigid offset to refine an alignment result; And secondly, in the fusion stage, based on the shape of the overlapped area of the aligned images, calculating pixel-level smoothing weights, wherein the pixel-level smoothing weights meet continuous boundaries and the sum of weights at the same position is 1, decomposing the images to different frequency layers, fusing the images layer by layer according to the pixel-level smoothing weights, and reconstructing to obtain a spliced image.
  2. 2. The unmanned aerial vehicle image-oriented high-fidelity stitching method of claim 1, wherein the alignment stage is used for achieving accurate image positioning through coarse-fine two-stage processing and is used for solving the problems of large parallax and non-rigid deformation.
  3. 3. The method and system for high-fidelity stitching for unmanned aerial vehicle images according to claim 1, wherein in the process of adaptively adjusting feature weights, key feature areas are highlighted and interference of irrelevant background areas is suppressed.
  4. 4. The method for high-fidelity stitching for unmanned aerial vehicle images according to claim 1, wherein the smoothing limitation imposed on the local offset is used for keeping the structure of rigid targets such as buildings and the like natural and avoiding distortion.
  5. 5. The method for high-fidelity stitching for unmanned aerial vehicle images of claim 1, wherein the generation of the pixel-level weight mask is performed without suture line detection, and the method is suitable for overlapping areas with arbitrary shapes.
  6. 6. The method for high-fidelity stitching for unmanned aerial vehicle images according to claim 1, wherein the multi-band fusion is used for eliminating stitching joint and ghost phenomena and realizing natural transition of images.
  7. 7. The unmanned aerial vehicle image-oriented high-fidelity splicing system is suitable for the unmanned aerial vehicle image-oriented high-fidelity splicing method according to any one of claims 1 to 6, and is characterized by comprising an image acquisition unit, a double-stage processing unit and an output display unit, wherein the image acquisition unit is used for acquiring a target area image in the flight process of an unmanned aerial vehicle and transmitting acquired image data to the double-stage processing unit, the double-stage processing unit comprises an alignment module and a fusion module, the alignment module is responsible for realizing global estimation, local correction, self-adaptive enhancement and geometric smoothness constraint functions, the fusion module is responsible for realizing weight mask generation and multiband fusion functions, and the output display unit is used for receiving the spliced image output by the double-stage processing unit and displaying or storing the spliced image.

Description

Unmanned aerial vehicle image-oriented high-fidelity stitching method and system Technical Field The invention relates to the technical field of unmanned aerial vehicles, in particular to a high-fidelity splicing method and system for unmanned aerial vehicle images. Background The unmanned aerial vehicle is an unmanned aerial vehicle which does not need to be controlled by a driver in boarding, can be controlled by radio remote control equipment or a self-provided program control device, integrates the multi-field technologies such as aerodynamics, electronic communication, automatic control and the like, can be divided into three categories of a consumption level, an industrial level and a military level according to application scenes, is mainly used for aerial photography and entertainment, can bear the tasks of agricultural plant protection, electric inspection, surveying and mapping exploration, logistics distribution and the like, can execute military operations such as reconnaissance, monitoring, striking and the like, has the characteristics of flexible maneuvering, low operation cost, adaptability to complex dangerous environments and the like, and is widely applied to a plurality of fields such as civilian life, industry, national defense and the like; In the actual use process of the existing unmanned aerial vehicle, the unmanned aerial vehicle is required to collect the picture images, but the shooting angle is unstable due to the fact that the unmanned aerial vehicle is easy to be interfered by air flow in the flight process, and a complex three-dimensional structure exists in a shooting scene, so that the problems of visual difference, distortion and irregular overlapping commonly exist among the collected images, the accurate alignment of a splicing technology is caused, an alignment error exists, and the reality of image splicing is affected; therefore, we propose a high-fidelity stitching method and system for unmanned aerial vehicle images. Disclosure of Invention The invention aims to provide a high-fidelity splicing method and system for unmanned aerial vehicle images, which solve the problems in the background technology. In order to achieve the purpose, the invention provides the following technical scheme that the high-fidelity splicing method for the unmanned aerial vehicle image comprises the following steps: The method comprises the steps of firstly, in an alignment stage, realizing coarse alignment by calculating integral transformation parameters between two images, and further estimating local non-rigid offset to refine an alignment result; And secondly, in the fusion stage, based on the shape of the overlapped area of the aligned images, calculating pixel-level smoothing weights, wherein the pixel-level smoothing weights meet continuous boundaries and the sum of weights at the same position is 1, decomposing the images to different frequency layers, fusing the images layer by layer according to the pixel-level smoothing weights, and reconstructing to obtain a spliced image. As a preferred embodiment of the present invention, the alignment stage achieves accurate image positioning by a "coarse-fine" two-stage process for handling large parallax and non-rigid deformation problems. As a preferred embodiment of the present invention, the adaptive adjustment of the feature weights highlights key feature areas and suppresses interference of irrelevant background areas. As a preferred embodiment of the present invention, the smoothing limitation imposed on the local offset is to keep the structure of the rigid object such as a building natural and avoid distortion. As a preferred embodiment of the present invention, the pixel-level weight mask is generated without performing suture line detection, and is adapted to an overlapping region of an arbitrary shape. As a preferred embodiment of the invention, the multi-band fusion is used for eliminating the phenomena of splicing seams and ghosts, and realizing natural transition of images. The invention further relates to a high-fidelity splicing system for the unmanned aerial vehicle image, which comprises an image acquisition unit, a double-stage processing unit and an output display unit, wherein the image acquisition unit is used for acquiring a target area image in the flight process of the unmanned aerial vehicle and transmitting acquired image data to the double-stage processing unit, the double-stage processing unit comprises an alignment module and a fusion module, the alignment module is responsible for realizing global estimation, local correction, self-adaptive enhancement and geometric smoothness constraint functions, the fusion module is responsible for realizing weight mask generation and multiband fusion functions, and the output display unit is used for receiving the spliced image output by the double-stage processing unit and displaying or storing the spliced image. Compared with the prior art, the invention has the follow