RGB空间平均颜色矢量法在颜色识别中的研究

RESEARCH ON APPLYING AVERAGE COLOUR VECTOR METHOD IN RGB SPACE TO COLOUR RECOGNITION

目前基于颜色的检索方法很多。引入平均颜色矢量,用平均颜色矢量方法对彩色图像中多个单种颜色目标进行识别。通过计算和分析图像像素与各类训练区的Euclidean距离及各颜色分量平均值和标准差,判断目标颜色相似性,结合Canny算子进行边缘检测,分割图像中的目标。该方法能对RGB值较为明显或颜色分布丰富且部分背景与目标颜色相近的图像进行精确的颜色识别,在番茄色选仿真实验中取得了令人满意的效果。

At present,there are many colour-based retrieval methods. In the paper we introduce the average colour vector and use this method to recognise the multiple monochromatic targets in colour image. By calculating and analysing the Euclidean distance between the image pixel and a variety of training districts as well as the average value and standard deviation of each colour component,we judge the similarity of targets colour,make edge detection in combination with Canny operator and segment the target in the image. This method can accurately recognise the colours of the image which RGB value is quite obvious or with rich colour distribution and part of the background to be close to the target colour. The method reaches satisfied effect in simulation experiment of tomato colour choosing