微小尺寸精度图像角点特征在线搜索仿真研究

Online Search Simulation of Corner Features of Tiny-size Accuracy Images

针对当前方法无法快速、准确地对微小尺寸精度图像角点特征在线搜索的问题,提出基于纹理结构引导的微小尺寸精度图像角点特征在线搜索方法。对含噪的微小尺寸精度图像进行小波分解,并估计噪声方差,引入噪声方差的修正因子,对噪声方差进行修正。根据修正结果经维纳滤波器对微小尺寸精度图像中含有的噪声进行去除,再采用小波系数对微小尺寸精度图像进行重构,实现微小尺寸精度图像增强。在微小尺寸精度图像角点特征在线搜索过程中引入结构张量,完成微小尺寸精度图像角点特征的搜索。利用张量结构特征定量分析微小尺寸精度图像角点特征在线搜索,得到微小尺寸精度图像角点特征信息填补优先级判断函数。将曲率引入微小尺寸精度图像角点特征在线搜索中,获取最适当的纹理合成模板,利用优先级判断函数和纹理合成模板获取微小尺寸精度图像角点特征在线搜索最终的搜索准确率。实验表明,上述方法微小尺寸精度图像角点特征在线搜索平均耗时为9μs,微小尺寸精度图像角点特征在线搜索后的准确性良好。所提方法分别在微小尺寸精度图像角点特征在线搜索耗时和微小尺寸精度图像准确性方面均有所保障,具有可实践性。

An online searching method of corner features for micro size precision images based on texture structure guidance was proposed. Wavelet transform was used to decompose the micro size precision image with noise. The noise variance was estimated,and the correction factor of noise variance was introduced to correct the noise variance. According to the corrected results,the noise in the micro size precision image was removed by Wiener filter,and then the wavelet coefficients were used to reconstruct the micro size precision image to realize the image enhancement of micro size precision. The structure tensor was introduced in the online search process to complete the search of the corner features of micro size precision images. The tensor structure feature was used to quantitatively analyze the corner features,so as to get the information filling priority judgment function. Then the most appropriate texture synthesis template was obtained. The final search accuracy was obtained by using the priority judgment function and the texture synthesis template. Simulation results show that the average time consuming of searching the corner features of the micro size precision image is 9μs,and its accuracy is good. The proposed method is guaranteed in the time consuming and the accuracy.