光照不均图像的自适应加权均值算法

 在结合矩形框加权均值算法的基础上,提出一种对光照不均图像进行二值化处理的自适应加权均值算法。对流量计图像中每个像素点分别计算阈值,利用和值表及自适应计算调整参数t,可有效地避免光照不均的影响,减少程序计算量,增强阈值自适应能力,提高流量检测速度及图像二值化处理效果。结果显示,在C++环境下,分别运用自适应加权均值算法和传统的基于小波包及Niblack 算法,对所采集的同一幅流量计光照不均图像进行二值化处理,自适应加权均值算法用时1.6 ms,基于小波包及Niblack 算法用时9.3 ms,且自适应加权均值算法处理效果明显优于传统算法。

A daptive weighted mean algorithm for uneven illuminated image

In order to obtain the field flowmeter data that can not be transmitted as electrical signals automatically, the binarization is necessary. The current method can not eliminate the influence of the uneven illumination in processing the flowmeter data. Therefore, an improved weighted average algorithm combined with a rectangular box is put forward. The method of the summed-area table is used to calculate the sum of the gray values in each rectangular area of the flowmeter image, which reduces the running time effectively. Meanwhile, the method of the class space combined with the cohesion is applied to automatically adjust the parameters. With the C++ programming and the method, the image processing time is 1.6 ms. For the same image, the image processing time by the Wavelet Packet and the Niblack Method is 9.3 ms. Therefore, the image processing speed is significantly improved by the method. The experiments show that the binarization of the unevenly illuminated image is improved with the method and a better image is obtained from the flowmeter data. Hence, the accuracy of the flowmeter is improved.