Registration and Fusion Method for Realizing Multispectral Pseudo-color Images

Abstract

The invention discloses a method for achieving registration and fusion of multi-spectral pseudo color images and belongs to the field of a spectral image processing technology. The method is achieved through an image registration and fusion software package which is arranged in a multi-spectral imaging device. The method includes that after the multi-spectral imaging device is powered on, the registration is performed on spectral images which have the same wave band in four imaging channels to obtain a registration center coordinate; translation processing is performed on four spectral images with different wave lengths in corresponding imaging channels with the registration center coordinate as the center to obtain a registration image; RGB (Red Green Blue) pseudo color fusion is performed on the image after the registration and the image is sent into a ground control station. The method for achieving the registration and fusion of the multi-spectral pseudo color images has the advantages of solving the problem that the multi-spectral images are difficult to be registered due to a large gray difference, improving the success rate of the registration, achieving the rapid registration of the multi-spectral images without manual intervention and being high in real-time performance and beneficial to engineering application.

Description

technical field [0001] The invention belongs to the technical field of spectral image processing, and mainly relates to a registration and fusion processing method of spectral images, in particular to a registration and fusion method of multi-spectral pseudo-color images. Background technique [0002] The spectral image is a narrow-band spectral grayscale image centered on a certain wavelength point and with a bandwidth of ±10nm or less obtained through optical spectroscopy. In the range of 0.4-1.0um visible light to near-infrared band, it can acquire spectral images of multiple wavelength points. Since the spectral distribution of different target energies is different, using this characteristic, different targets can be distinguished in the spectral image. Therefore, it has been widely used in many fields such as biomedicine, agricultural and forestry geological remote sensing detection, and military target reconnaissance. [0003] Multi-spectral image pseudo-color fusio...

AI Technical Summary

Problems solved by technology

[0006] Chinese patent ZL2008100645595.5 discloses a multispectral image registration method based on small components and noise components. The method first performs scale transformation, region selection, and control point selection preprocessing on the multispectral image, and then uses the traditional grayscale-based And the feature-based method uses a certain spectral image as a reference image to initially register other images one by one, and then uses the analysis method of small components and noise components to perform repeated accurate registration, and its registration accuracy is significantly better than that of traditional methods. The method based on grayscale and features, but its calculation process is complex, the amount of calculation is large, the real-time performance is poor, it is difficult to implement in engineering, and it cannot meet the requirements of actual tactics.

Images