公开(公告)号：US09082019B2

公开(公告)日：2015-07-14

申请号：US13918513

申请日：2013-06-14

Applicant: NATIONAL YUNLIN UNIVERSITY OF SCIENCE AND TECHNOLOGY

Inventor： Ming-Hwa Sheu ,  Wen-Kai Tsai ,  Jian-Hui Chen

IPC: G06K9/00

CPC classification number: G06K9/00771 ,  G06K9/4638 ,  G06T7/11 ,  G06T7/194 ,  G06T2207/10024 ,  G06T2207/20224 ,  G06T2207/30232

Abstract: A method of establishing an adjustable-block background model for detecting a real-time image object is provided to obtain a surveillance image by a surveillance apparatus. The surveillance image has a plurality of pixels. The method includes steps of: segmenting the surveillance image into a plurality of blocks each having a first pixel and at least one second pixel; defining the first pixel as a major color and comparing the first pixel with the at least one second pixel to determine a number and color information of the major color in the block; merging the blocks having the same major color into a large block to obtain a block background model; and performing image comparison to identify a moving object image. With the establishment of the block background model, a required memory space is effectively reduced while outstanding image display performance is still maintained.

Abstract translation: 提供一种建立用于检测实时图像对象的可调块背景模型的方法，以通过监视装置获得监视图像。 监视图像具有多个像素。 该方法包括以下步骤：将监视图像分割为多个块，每个块具有第一像素和至少一个第二像素; 将所述第一像素定义为主色，并且将所述第一像素与所述至少一个第二像素进行比较，以确定所述块中的主色的数量和颜色信息; 将具有相同主色的块合并成大块以获得块背景模型; 并执行图像比较以识别移动物体图像。 随着块背景模型的建立，有效降低了所需的存储空间，同时保持了出色的图像显示性能。