Multi - Image Fusion Algorithm And Its Application

Image fusion technology is to make full use of multiple source images to generate fusion image by comprehensive processing. The fusion image minimizes the redundant information of source image, and be more complete and reliable than any source images, improving image information. Multi-sensor image fusion technology can be widely used in large depth of field imaging of digital cameras, comprehensive display of the medical multiple pattern images, the earth remote sensing image classification recognition, robot vision, virtual reality,3D image reconstruction, defense security checks, and other fields.According to the different source of images, the multi-source image fusion can be roughly divided into:multi-focus image fusion, multi-sensor image fusion, the multi-source remote sensing image fusion and time series (dynamic) image fusion. As an important branch of image fusion technology, multi-focus image fusion technology is mainly used in image processing, making the objects which have different imaging distance can be clearly in an image. Multi-focus image fusion technology can use image information effectively and improve the reliability of the detection system to identify the target, and laid a good foundation of processing such as feature extraction, image recognition.Through the probe and research into quantification, compression and fusion of multiple images and multi-focus image fusion, this paper proposed new algorithm of image fusion.For the compression and fusion of multiple images, this paper firstly proceed block DCT of source images, secondly each sub-block select coefficient according to "Z" font. And then, using the adaptive quantization rules on the basis of assuring the quality of reconstructed image to proceed maximum compression. At last, use some fusion rules to get final fusion image. The fused image has some concealment, using the correct key can extract the image information. Experiment results show that this algorithm can fuse abundant image information, and can compress and fuse different sizes and grayscale images. The reconstructed image achieves a satisfactory result.For multi-focus image fusion, based on the analysis of the characteristics of wavelet transform and the basic theory of mathematical morphology, this paper combined with a variety of general image evaluation index, proposing a fusion method which applies to multi-focus images, and proved the validity of this method. For the fusion of low-lighting images, use sobel operator and the theory of morphology to realize the fusion in the wavelet domain. The experiment results show that the fusion image entropy, spatial frequency and average gradient are obviously enhanced compared to before-fusion. The three evaluation parameters show that the fusion image contains more information and be more distinct, and the fusion effect is satisfactory.