A target detection method for vehicle vision system

A target detection and vision system technology, which is applied in the target detection field of vehicle vision systems, can solve the problems of large environmental changes and low performance accuracy, and achieve the effect of shortening the target detection time and improving the target detection performance

CN108256444BActive Publication Date: 2020-10-09BEIJING HUAHANG RADIO MEASUREMENT & RES INST
2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Publication Date
2020-10-09

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention relates to a target detection method for a vehicle-mounted vision system. According to the posture presented by the target in the visual image, a training sample library is constructed, and an approximate channel feature pyramid is extracted from the training sample library samples; an algorithm based on soft-cascade is used to train multiple Scale cascade classifier model; use the trained multi-scale cascade classifier for target detection on the loaded image to be detected; perform window merging and screening on the obtained target detection windows to obtain the final target detection output. The present invention is applicable to various target detections in the application environment of the vehicle-mounted vision system, such as vehicles, pedestrians, bicycles, etc.; the multi-model target detection algorithm based on the approximate channel feature pyramid greatly improves the target detection performance, and at the same time greatly shortens the Object detection time; removal of occluded objects is avoided by weighting corrections to the confidence of overlapping detection windows.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention relates to the technical field of image processing, in particular to a target detection method for a vehicle-mounted vision system. Background technique

[0002] The vehicle vision system has a very broad application prospect in the fields of intelligent transportation, automatic driving, assisted driving and intelligent robots. From a technical point of view, the target detection method based on vehicle vision mainly includes feature extraction, classifier design, target detection and detection result suppression.

[0003] The on-vehicle vision system captures the natural scene images in front of the road through the front-view camera installed on the driving vehicle, and inputs them into the target detection module in the system to complete the processing. Due to the complexity and changeability of natural scenes, it is more challenging than target detection in unnatural scenes. The difficulties mainly include susceptibility to illumin...

Examples

Embodiment Construction

[0040] Preferred embodiments of the present invention will be specifically described below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of the application and are used together with the embodiments of the present invention to explain the principles of the present invention.

[0041] A specific embodiment of the present invention discloses a target detection method for a vehicle-mounted vision system; figure 1 shown, including the following steps:

[0042] Step S1, divide the target into N intervals according to the posture presented by the target in the visual image, construct a training sample library composed of N training subsets, and extract an approximate channel feature pyramid from the training sample library samples;

[0043] Step S101, dividing the training sample library composed of multi-angle target images into N training sample subsets;

[0044] Divide the view angle presented by the target into N intervals at...