基于改进Canny算子的光纤面板暗影检测

Optical fiber panel shadow detection based on the improved Canny operator

暗影作为光纤面板的缺陷之一,对其成像质量有很大的影响。首先利用最优算子Canny算子实现暗影的检测,然后通过分析其结果,提出了利用小波去噪的方法改进Canny算子。并提出在小波去噪的始末增加对图像取对数、指数的操作,从而同时去除加性噪声和乘性噪声。实现了在VC++开发环境下将改进的小波去噪算法嵌入到Canny算子中,代替其本身内置的高斯滤波器。实验结果表明,改进后Canny算子检测出的暗影边缘结果优于改进前的Canny算子,且避免了手动设置参数,很大程度上降低了程序的复杂性。

Shadow is one kind of defects of optical fiber panel （OFP） which can cause serious effect to the quality of the transmitted image. This paper realized the shadow detection by the Canny operator firstly. Through analysing the detection results, one method using wavelet denoise to improve Canny operator was proposed. And by adding logarithm operation to the image at the beginning of the wavelet denoise and exponent operation at the ending of the wavelet denoise, additive noise and multiplicative noise were eliminated simultaneously. In this paper, this improved wavelet denoise arithmetic was embedded in the Canny operator to replace its built-in Gaussian filter. And this method was realized in VC＋＋. The experimental results indicated that the result detected by improved Canny operator was much better than the result detected by original Canny operator. And the improved Canny operator could avoid setting parameter manually, reduced the complexity of the program obviously.