激光与光电子学进展, 2020, 57 (12): 121003, 网络出版: 2020-06-03
基于方向梯度直方图和灰度共生矩阵混合特征的金文图像识别
Recognition of Bronze Inscriptions Image Based on Mixed Features of Histogram of Oriented Gradient and Gray Level Co-Occurrence Matrix
摘要
金文图像识别的关键步骤是提取金文的图像特征,针对金文的形态特点,提出了一种基于方向梯度直方图(HOG)和灰度共生矩阵(GLCM)的金文图像识别算法。使用双边滤波对金文图像进行预处理,针对金文的结构特征和局部纹理特征,提取其HOG特征和GLCM特征并将二者进行融合。用融合后的特征作为样本训练支持向量机分类器,用训练后的模型识别金文图像。实验结果表明,该算法相比基于HOG特征的算法,分类准确率提高了19.47个百分点,能更好地提取金文的图像特征,提高识别的准确率。
Abstract
The key step in the process of image recognition of bronze inscriptions is to extract their image features. According to the morphological characteristics of bronze inscriptions, an inscription recognition algorithm based on histogram of oriented gradient (HOG) and gray level co-occurrence matrix (GLCM) is proposed. Using bilateral filtering to preprocess the inscription images, the HOG features and GLCM features are extracted and fused aiming at the structure and local texture features of the bronze inscriptions. The fused features are used as samples to train support vector machine classifier, and the trained model is used to identify the bronze inscription image. Experimental results show that compared with algorithms based on HOG feature, this algorithm improves the classification accuracy by 19.47 percentage points. and can better extract the image features of bronze inscriptions and improve the accuracy of recognition.
赵若晴, 王慧琴, 王可, 王展, 刘文腾. 基于方向梯度直方图和灰度共生矩阵混合特征的金文图像识别[J]. 激光与光电子学进展, 2020, 57(12): 121003. 赵若晴, 王慧琴, 王可, 王展, 刘文腾. Recognition of Bronze Inscriptions Image Based on Mixed Features of Histogram of Oriented Gradient and Gray Level Co-Occurrence Matrix[J]. Laser & Optoelectronics Progress, 2020, 57(12): 121003.