基于小波变换与Retinex的电路板红外图像增强技术

摘要

电路板红外图像发热芯片区域的增强是红外图像故障检测系统中一项重要研究内容。针对电路板某些芯片的发热量小，芯片区域红外图像与背景差异弱的特点，为提高故障检测效率，设计了一种融合小波变换与改进的多尺度Retinex红外图像增强算法。首先，通过小波变换获得不同频率的子带图像；然后，利用多尺度Retinex算法对低频子带图像进行增强处理；对于高频子带图像，利用可变阈值去噪，引入图像清晰度参数，依据参数值对高频图像进行适度增强。最后，对处理后的高、低频图像进行小波逆变换与重构，实现电路板红外图像增强。对增强后的红外图像进行的定量以及定性评价表明：与单一的利用直方图均衡算法、小波变换法以及多尺度Retinex增强算法相比，本文算法改善了某些发热芯片区域红外图像对比度低且细节模糊问题，抑制了噪声，提升了电路板红外图像整体视觉效果。%The chip area image enhancement of the circuit board is an important part of the infrared image fault detection system. Because some chip of circuit board has low calorific value, there is no obvious difference between the chip area infrared image and the background area infrared image. To improve the fault detection efficiency, a new infrared image enhancement method based on the wavelet transform and the improved multi-scale Retinex is designed in this paper. Firstly, the wavelet transform is used to obtain sub-band images of different frequency. Then, for low frequency sub-band image, the multi-scale Retinex is used, and for high frequency sub-band image, the variable threshold is applied to de-noise. At last, for the low and high frequency image, the image reconstruction is taken to achieve image fusion. The experiment results show that the method can suppress noise while preserving edge and detail elements, and enhance visual effect effectively.