High-performance image feature extraction and matching method and system and storage medium

Abstract

The invention discloses a high-performance image feature extraction and matching method. The method comprises the steps that a panoramic image of an automobile is acquired; performing inverse perspective transformation on the panoramic image and then establishing an image coordinate system; establishing a circle with a preset size in the panoramic image range; recording the gradient amplitude of each pixel in the circle; projecting pixels in the circle to the circle, and recording image features; and performing feature comparison on the image features of the circle and the material features. The invention further discloses an image feature extraction and matching system. The system realizes the method. The invention further discloses a storage medium, and when a program stored in the medium runs, a carrier of the program runs to achieve the method. When the material features are matched with the image features, images at any angle can be matched; meanwhile, the material characteristicsare simple, the calculated amount is small and the matching precision is high.

Description

technical field [0001] The invention relates to the field of image analysis methods and systems, in particular to a high-performance image feature extraction and matching method, system and storage medium. Background technique [0002] In the field of driving assistance technology, the traditional map building method mainly relies on information such as wheel speed pulses and steering wheel angles on the car body to estimate the vehicle's pose using inertial navigation algorithms, and at the same time uses ultrasonic radar to mark the surrounding obstacle information on the map. . Due to the existence of mechanical errors, the traditional map will inevitably produce errors during the long-term establishment process, so it is necessary to use other information feedback to assist in updating the map, and the panoramic vision system on the vehicle body can provide an efficient and visual solution; local electronic map The establishment of is of great significance for the navig...

AI Technical Summary

Problems solved by technology

The features generally select basic features such as image invariant features, texture features, and corner features. Since the ground features are mostly smooth lines, the corner features are not obvious; while using other feature statistics requires a large amount of calculation, and the physical interpretability is not good. powerful

Images