Image segmentation based car window extraction method and system

Abstract

The invention discloses an image segmentation based car window extraction method and system. The method comprises the steps of normalizing an obtained vehicle image to obtain a normalized image; performing undirected graph based image segmentation processing on the normalized image; according to a distance between a mass center of each region and a mass center reference point, sorting regions in the upper half parts of the segmented images, then adding the regions into a cache image in sequence according to a sorting result, calculating the regional boundary fluency of the cache image in real time by adopting a first-order difference of a Freeman chain code, and extracting candidate car window regions according to the regional boundary fluency of the cache image; and according to a set car window region judgment basis, screening out a final car window region from the candidate car window regions. According to the method and system, the undirected graph based image segmentation method is adopted, so that the anti-jamming capability is good and the robustness is high; and meanwhile, the extraction effect and accuracy of the car window region are ensured. The method and system can be widely applied to the field of image processing.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a method and system for extracting car windows based on graph segmentation. Background technique [0002] In recent years, intelligent transportation system has achieved rapid development and wide application, and traffic safety is an important part of it. Safe driving is a key part of traffic safety, so the safety monitoring of drivers is particularly important. Driver safety monitoring includes monitoring whether the driver is wearing a seat belt, making phone calls while driving, and fatigue driving. In addition, the detection of annual inspection marks and environmental protection marks is also an important part of traffic vehicle management. The accurate extraction of the window area is the premise to solve the above problems, and it is also an important content in the field of traffic image processing and pattern recognition. [0003] Existing vehicle window extraction me...

AI Technical Summary

Problems solved by technology

[0004] However, the edge-based window extraction method has poor anti-interference ability. When the light is poor or the color of the car is dark, the method will fail because the detected edges are often less; in addition, when the vehicle in the image deviates from Larger angles or different window shapes for different car models can also defeat the effect of this method due to the difficulty of finding a straight line at a specific angle in the resulting edge

However, in the color-based window extraction method, in the environment with strong light, when the car body is stained or painted with multiple colors, the chromaticity difference of the car body color will also be large, resulting in the extraction effect of this method is not good enough

Images