摘要
局部二值模式(LBP)在纹理特征提取时,易受光照、旋转、噪声等复杂条件的影响.本文定义一种新型自适应局部二值模式,通过考虑模式的均匀度和相似度,来实现纹理模式分类和特征值计算.结合差分运算,分别在差分二值矩阵和差分绝对值矩阵上计算自适应纹理特征,并将两部分特征值连接成一个空域增强的特征向量,采用最近邻分类器完成图像分类识别.实验结果表明,该算法在复杂条件下具有更好的识别效果.
The paper define a new adaptive local binary pattern, abbreviate ALBP, which employs the uniformity and similarity of texture patterns to classify different patterns and then re-label them to enhance the robustness under different illuminant, noise, scaling, rotation, and translation. Combing with differential operation, a local region is represented not only by its local difference sign but also through magnitude matrix. Finally, the two part eigenvalues are concatenated into an enhancement feature vectors, and used for texture classification. The classification was conducted using a nearest neighborhood classifier in the computed feature space with Chi-square as a dissimilarity measure. Experiments and comparisons show that the proposed method has better result under different rotation, illuminant and noise conditions.
出处
《四川大学学报(自然科学版)》
CAS
CSCD
北大核心
2012年第5期995-1002,共8页
Journal of Sichuan University(Natural Science Edition)
基金
四川省科技厅苗子工程项目(2011-053)
可视化计算与虚拟现实四川省重点实验室课题(PJ201103)
四川师范大学研究生科研创新基金项目(2011-022)