结合HSL模型与傅里叶描述子的三维彩色物体识别

Recognizing a colored object based on an HSL model and Fourier descriptors

目标物识别是机器人导航中重要的一步,现存的方法大多对于场景中的彩色物体仅采用颜色分割,或者转化为灰度图进行识别,不能满足彩色物体识别的需要.提出在基于HSL模型颜色分割的基础上,结合傅里叶算子对轮廓特征识别的优势,先用三维目标各个角度成像的傅里叶描述子建立分类器,再对三维物体的二维成像进行轮廓特征识别,并在颜色分割的过程中采用了快速算法.实验表明,物体测试集的识别率达到了73.3%,可应用于对实时性要求比较高的彩色物体智能识别系统.

Object recognition is an important step for achieving robot navigation. There are many approaches in this domain to recognize a colored object. Most of them segment an image by color to determine the target, or convert the color image to grey before recognizing it. This study focused both on color and shape information. First, the im- age was segmented based on an HSL model and flood fill algorithm; secondly, by combination with the Fourier operator that has advantage in contour feature recognition, build a classifier by Fourier descriptor to discriminate a three-dimensional object from every angle, and then identify the two-dimensional image＇ s outline feature of a three- dimensional object, and use a fast algorithm during color segmentation. The experiments show that the average recognition rate reaches 73.3% on the testing set. The method proposed in this paper can be applied to colored object intelligent recognition system which has higher real-time requirement.