Image Fusion And Image Fusion Evaluation Based On Contourlet Transform

Image fusion technology refers to the process, which captures images indifferent angles at the same time or in the same angle at different times for the sameobject or target by the multi-channel sources. The unique characteristics of theseimages has been got by a characteristic value extraction technology at all angles.Finally, a more accurate and comprehensive picture was acquired by a certainalgorithm, which removes redundant information and makes these informationcomplementary. These operations make characteristics and accuracy of the finalfused image be better than the previous image. The target scene is also restored moreaccurately. These fused images can be used for the analysis, observation or othertasks in the future.Contourlet transform image fusion is used as a transformation methodcommonly which improved on the basis of the wavelet transform. It retains the ideaof multi-scale in wavelet transform and achieves the decomposition at any directionon any scale. So contourlet transform descripts the details of the images to be fusedbetter, for example, the information of contour and texture. It also optimizes imageeffects after the wavelet transform. PCA transform method is a typical and traditionalimage fusion. It reduces the image data by mensionality reduction to generate aplurality of uncorrelated components. So that most of the important information ofthe source image data can be stored in the first component.In this paper, an image fusion method based on contourlet transform and PCAtransform was proposed. Three sets of images were used to verify the feasibility. Forthe comparison of different image algorithm, we adopted three methods which wereGradient Pyramid, FSD Pyramid and Average Pyramid. In order to value the prosand cons of the fused image, we used four methods, which were Standard Deviation, Entropy, Mutual Information and SSIM. Then a comprehensive assessment of imagequality was got combined with subjective evaluation results.By the analysis of image classification and evaluation indicators selection, wealso present feature-level image fusion performance evaluation model based oncontourlet transform. On the basis of the studying and summarizing the single-scalefeature-level and multi-scale feature-level image fusion assessment method, weproposed a novel multi-scale, multi-directional image fusion performance evaluationmeasure. In this measure, we integrate the contourlet transform into the fusionevaluation, then, the similarity between source images and fused images arecalculated in terms of edge and texture information. The better results are obtainedafter compared with the subjective evaluation results, as well as other evaluationmeasures. Experimental results show that the measure is reliable. Experimental datashows that the evaluation result of the contourlet transform after feature extraction isbetter than the effect of other comparative evaluations.