摘要
当前图像缩放方法耗时长、缩放完成后图像质量得不到保障,提出基于多小波逆变换的数字图像曲面缩放方法。利用Harris角点检测法对数字图像中重要角点进行检测的过程中,设置角点响应函数,通过响应函数值与选定门限的比较,确定图像中的重要角点,并将重要角点输出。基于数字图像角点检测结果,对原图像进行小波分解,并根据三次样条插值的方式实现图像中高频子带系数的放大。利用自适应阈值法对放大之后的系数进行处理,使用非线性增强函数转变原图像各个尺度上的细节分量间对比度,将所得结果引入多小波逆变换中,得到缩放后的数字图像。仿真结果表明,基于多小波逆变换的数字图像曲面缩放方法运行耗时短、缩放后图像分辨率高,与当前相关方法相比更具科学性和可实践性。
At present, the image scaling method is time-consuming. Meanwhile, the quality of image cannot be guaranteed. Therefore, a digital image surface scaling method based on multi-wavelet inverse transform was presented. Firstly, Harris corner detection method was used to detect the important corners in digital image. During this process, the corner response function was set. By comparing the response function with the selected threshold, important corners in image were determined, and then important corners were output. Based on the digital image corner detection result, the original image was decomposed by wavelet. After that, the high-frequency sub-band coefficients were amplified by the way of cubic spline interpolation. In addition, the adaptive threshold method was used to process the coefficients after amplification. Moreover, the non-linear enhancement function was used to change the contrast between the detail components at different scales of original image. Finally, the result was introduced into the multi-wavelet inverse transform to obtain the scaled digital image. Simulation results show that the digital image surface scaling method based on multi-wavelet inverse transform occupies less time. Meanwhile, this method has high image resolution, which is more scientific and practical than that of current methods.
作者
丁永胜
DING Yong-sheng(School of science,Qiqihar University,Qigihar Heilongjiang 161006,China)
出处
《计算机仿真》
北大核心
2019年第11期335-338,共4页
Computer Simulation
基金
黑龙江省教育厅科学技术项目项目编号(135209247)
关键词
多小波逆变换
数字图像
曲面缩放
Multi-wavelet inverse transform
Digital image
Curved surface scaling
