为克服基于码本模型的目标检测算法在光照变化条件下对检测结果产生的影响，提出了一种融合卡尔曼滤波思想的改进码本算法.该方法在计算图像像素亮度变化时选用了 YUV 颜色空间，使空间坐标轴与亮度变化方向一致，使亮度变化的计算量平均下降了44.9％.同时在建立码本模型时为每个像素初始化一个卡尔曼滤波器，该滤波器利用前后两帧图像亮度信息预测与修正当前像素值，对光照变化有较好的适应性.仿真实验结果表明，该算法与 YUV 码本模型、RGB 码本模型以及 GMM 算法相比在亮度变化的条件下对噪声的抑制作用更强，体现出更好的自适应性.%In order to overcome the influence of illumination changes on detecting results by using co-debook-based object detection algorithm,an improved codebook algorithm that fuses Kalman filtering is proposed in this paper.The algorithm calculates the pixel brightness change in YUV colour space, which corresponds brightness changes with spatial coordinates and makes the calculation of brightness process fallen by an average of 44.9%.Meanwhile,it initializes a Kalman filter for each pixel at the time when the codebook model is established and the filter uses two successive image’s brightness in-formation to predict and correct current pixel value.This method has good adaptability on brightness changes.The experimental results show that this algorithm reflects a better adaptability under the condition of brightness changes,compared with ordinary YVU codebook model,RGB codebook model and GMM method.
展开▼
