An improved FREAK feature point matching image stabilization method suitable for tunnel environmental characteristics

Abstract

The invention relates to the technical field of traffic image processing, and discloses an improved FREAK feature point matching image stabilization method suitable for tunnel environment characteristics. The method comprises the following steps: 1) obtaining video image and establishing reference frame; 2) performing multi-scale Gaussian fitting on the video image; 3) dividing the fitting video image into a plurality of sub-region window; 4) carrying out LPQ feature weighted gray-scale projection on each sub-region window to obtain a plurality of row projection curve and column projection curves; 5) obtaining the row projection curve and the column projection curve according to in the step 4), estimating a shake vector, and correcting the current image frame by using the shake vector. Beginning from that actual environment of the expressway tunnel, the fuzzy robust LPQ feature weighting is used to increase the gray difference between pixels of the blurred image, the current frame is modified by multi-scale Gaussian estimation and the local difference between the projection curves of the current frame and the background frame is changed. Then the accuracy of global jitter vector estimation is improved based on the idea of multi-subregion window. Finally, an improved FREAK feature point matching image stabilization method is proposed, which is suitable for the characteristics oftunnel environment.

Description

technical field [0001] The invention relates to the technical field of traffic image processing, in particular to an image stabilization processing method for expressway tunnel video. Background technique [0002] Accurate extraction of traffic targets in tunnels is the key to abnormal event detection in tunnels. However, in the tunnel scene, there is vibration in the monitoring equipment, which makes the extracted vehicle target deformed, and the pedestrian target is connected to the background, which seriously interferes with the effective extraction of traffic targets. At the same time, the image quality of the tunnel image is blurred, vehicle lights and other interferences also increase the difficulty of tunnel shake video stabilization, resulting in the general effect of traditional image stabilization methods and not strong pertinence. Therefore, it is of great theoretical and practical significance to study the jitter video stabilization method for the tunnel environ...

AI Technical Summary

Problems solved by technology

Among them, for the image stabilization method based on edge-like feature matching, such as the "Shipborne Image Stabilization Method Based on Sea-sky Boundary Detection" (CN: 103514587B) applied by the Beijing Institute of Environmental Characteristics, this class is based on the Canny operator to detect edge lines, This type of method requires a clear edge line as the image stabilization benchmark, and the fuzzy tunnel image cannot extract continuous edge lines, so it is difficult to apply this type of method to achieve image stabilization; for image stabilization methods based on corner point feature matching, such as Dalian University of Technology "An embedded foggy real-time video image stabilization method" (CN: 105976330A) applied by the university. The feature description method of this type of method is generally represented by the feature point itself or the gray value of the pixel in the neighborhood. Adaptability to interference, image stabilization accuracy is low under interference; for image stabilization methods based on floating-point feature matching, such as "An electronic image stabilization method and system based on SIFT feature matching and VFC algorithm" applied by Wuhan Engineering University "(CN: 105306785A) and "Video Image Stabilization Method Based on SURF and Fuzzy Clustering" applied by Jiangnan University (CN: 106550173A). Compared with the description methods of the first two types of features, this type of features has a certain degree of scale scaling, rotation, blurring, and illumination changes. However, the image stabilization method based on this type of feature is time-consuming and difficult to guarantee real-time performance; for the image stabilization method based on binary feature matching, such as the "full-frame electronic stabilization based on feature matching" applied by Beijing Institute of Technology "Image method" (CN: 105872345A) and "An Electronic Image Stabilization Method for Video Cameras Based on Robust Feature Points" (CN: 107343145A) applied by Shanghai Institute of Technical Physics, Chinese Academy of Sciences. In the feature matching link between the reference frame and the current frame, the 0-1 binary feature only needs a simple Hamming (Hamming distance) distance calculation to quickly complete the feature In the pairing process, these binary features can increase the speed by ten to one hundred times under the premise of obtaining less than SIFT (ScaleInvariant Feature Transform) and SURF (Speeded-Up RobustFeatures) image stabilization accuracy

Images