立体视觉下连续翻页动画效果增强方法仿真

Simulation of Continuous Page Turning Animation Enhancement Method under Stereo Vision

需要研究一种连续翻页动画效果增强方法,能够增强翻页动画图像的色彩保留度,降低失真度,增强连续翻页动画在立体视觉中的实际应用效果。针对当前方法在实现过程中由于噪音影响导致连续翻页动画亮度变化不够均匀,信息失真度较高等问题,提出一种基于Huber-MRF模型的连续翻页动画效果增强方法,通过利用NLN算法计算相似性权值,运用正交阵序列向量构建的空间对翻页动画图像进行分解,经过迭代后获取去噪后的翻页动画图像。针对去噪后的翻页动画图像,计算构建MAP模型的能量函数、数据项和平滑项,对相邻像素集合阈值,同时对Huber函数线性局部斜率进行调节,获取增强后的连续翻页动画效果图像。仿真结果证明,所提方法能够均衡动画亮度,降低失真度,增强翻页动画模糊系数、标准差异系数、信息熵和边缘保持度,在实际应用中具有一定的现实意义。

In this paper, we focus on a method to enhance continuous page turning animation effect based on Huber-MRF model. At first, we used NLN algorithm to calculate the similarity weight. Then, we used the space constructed by sequence vectors of orthogonal matrix to decompose the page turning animation image and obtain the denoised image of page turning animation after iteration. Moreover, we calculated the energy function, data item and smooth item of MAP model, and then we set threshold for adjacent pixels. Simultaneously, we adjusted the linear local slope of Huber function. Finally, we obtained the continuous page turning animation image after the enhancement. Simulation results show that the proposed method can balance the brightness of animation and ensure the degree of distortion. Meanwhile, this method can enhance the fuzzy coefficient, the standard difference coefficient, the information entropy and the edge preserving degree, which has realistic meaning in practical application.