针对公路上成像设备直接获取得到的图像分辨率普遍较低的问题,采用基于稀疏表示的超分辨方法提高图像的质量.选取一定数量的高分辨率图像和低分辨率图像作为训练集分别用于构建针对车牌图像的高分辨率字典和低分辨率字典,从而提高了稀疏表示的性能;分析了字典原子数目及正则化参数对图像重建效果影响,给出了字典原子数目及正则化参数较理想的值.通过车牌图像的对比实验验证了该超分辨方法的有效性.%According to the low resolution of available license plate images , it uses super -resolution based on sparse representation to enhance quality of the image with low resolution , and picks large amounts of plate num-ber images to joint train high and low resolution dictionaries .This improves the performance of sparse representa-tion algorithm .Taking a set of contrast experiments , it proves the effectiveness of this algorithm efficiently , ana-lyzes the influences of dictionary size and regularization parameter on the reconstruction result .
展开▼
