基于正则化的多光谱图像二值化处理

为了进一步提升传统多光谱图像二值化处理方法的处理速度和抗噪性能,提出基于正则化的多光谱图像二值化处理方法。将正则化约束引入到多光谱图像去噪模型中,对现有的多光谱去噪模型进行改进。并利用正则化框架中的数据正则项对图像的噪声机制以及图像的先验信息进行建模,以实现多光谱图像去噪处理。根据去噪结果采用约束能量最小化方法获取多光谱图像的边缘信息,将最佳全局阈值法和局部阈值自适应方法在边缘信息的基础上相结合,实现对多光谱图像的二值化处理。仿真结果表明,所设计方法具有较强的抗噪能力和较快的处理速度,并且经处理后的图像分辨率较高,充分验证了上述方法的有效性。

Binarization of Multispectral Images Based on Regularization

In order to further accelerate the processing and improve the anti-noise performance of traditional methods, a method to binarize the multi-spectral image based on regularization was proposed. Firstly, regularization constraints were introduced into the multispectral image denoising model to improve the existing model. Then, the noise mechanism and prior information of image was modeled by the data regularization items in regularization framework, so as to realize the noise reduction of multi-spectral images. According to the denoising result, the constrained energy minimization method was used to obtain the edge information of multispectral image. Moreover, the optimal global threshold method and the local threshold adaptive method were combined on the basis of edge information. Finally, the binarization processing of multispectral image was realized. Simulation results show that the proposed method has strong anti-noise ability and fast processing speed, and the image resolution after processing is high. This simulation fully verifies the effectiveness of the proposed method.