发明名称 Method and apparatus for identifying a possible collision object
摘要 An imaging unit is arranged in a motor vehicle. The imaging unit is designed for providing, as a function of an image acquired by it, a digital source image of a predefined image size. A first intermediate image of a predefined first intermediate image size is generated by reducing a resolution of the source image for the sake of reducing pixels. Furthermore, a second intermediate image of the predefined image size is generated such that it comprises the first intermediate image. The second intermediate image is analyzed by a predefined detector in order to examine whether an object of a predefined object category is situated in the second intermediate image, the detector being designed for analyzing a predefined image detail and for detecting an object of a predefined object category of a predefined object size range.
申请公布号 US9305221(B2) 申请公布日期 2016.04.05
申请号 US201314082721 申请日期 2013.11.18
申请人 Bayerische Motoren Werke Aktiengesellschaft 发明人 Orecher Sebastian
分类号 G06K9/00;G08G1/16 主分类号 G06K9/00
代理机构 Crowell & Moring LLP 代理人 Crowell & Moring LLP
主权项 1. A method for detecting a potential collision object in an image, the method comprising the acts of: acquiring, by a camera arranged in a motor vehicle, the image; providing, by the camera, a digital source image of a predefined image size as a function of the acquired image; generating a first intermediate image of a predefined first intermediate image size by reducing a resolution of the digital source image in a sense of reducing pixels; generating a second intermediate image of the predefined image size based on the first intermediate image, the generated second intermediate image including both the first intermediate image and additional pixels not included in the first intermediate image; analyzing, via a detector, the second intermediate image to determine whether an object of a predefined object category is located in the second intermediate image, wherein the detector analyzes a predefined image detail and detects the object of the predefined object category of a predefined object size range, wherein a first pixel value is determined as a function of pixel values of the digital source image or of the first intermediate image, and the additional pixels of the second intermediate image, which are not included within the first intermediate image, have the first pixel value.
地址 Munich DE