基于自适应纹理复杂度的仿生视觉导航方法研究

针对稠密光流在低纹理复杂度时精度较低的问题,提出了一种自适应纹理复杂度的稠密光流优化方法,以提升光流导航精度。根据三种不同大气条件下三种不同图像模糊程度的图像光流精度与纹理复杂度的统计图,推断稠密光流的精度与图像的纹理复杂度呈线性关系。通过建立图像纹理复杂度和稠密光流精度之间的直接联系,利用灰度共生矩阵的对比度参数评价图像纹理复杂度,采用最小二乘法拟合图像纹理复杂度和光流真值优化系数的函数关系,获得自适应纹理复杂度的稠密光流优化模型。基于该优化模型设计了仿真实验,实验结果表明,基于该模型可有效提升稠密光流在低纹理复杂度时的计算精度。

Bionic Visual Navigation Method Based on Adaptive Texture Complexity

Aiming at the problem of low accuracy of dense optical flow at low texture complexity, this paper proposes a dense optical flow optimization method based on adaptive texture complexity to improve the accuracy of optical flow navigation. According to the statistical graphs of image optical flow accuracy and texture complexity in three different image blur levels as well as under three different atmospheric conditions, it is inferred that the accuracy of dense optical flow has a linear relationship with the texture complexity of the image. By establishing a direct relationship between image texture complexity and dense optical flow accuracy, the contrast of gray level co-occurrence matrix is used to evaluate the texture complexity of the image, and the least square method is used to fit the function relationship to obtain a dense optical flow optimization model with adaptive texture complexity. Based on the optimization model, the simulation experiments are designed. The results show that the calculation accuracy of dense optical flow in low texture complexity can be effectively improved based on the model.