Research Of Range-gated Active-passive Image Fusion At Pixel-level

Active and passive images obtained by imaging laser radar have their owncharacteristics. Active intensity images reflect laser reflectivity of the target, while rangeimages reflect distance information of it. The laser power is limited, so the edgeinformation of the target by laser imaging is poor; for passive images, the edge of thetarget is clearer, and contains more complete intensity information of the target, but thereis no distance information, and moreover, the target and background were integrated,which is not conducive to identify the target automatically. In this paper, the fusion ofrange-gated laser active and passive images was analyzed, simulated and experimentalverified. Data fusion of lidar image and passive optical image is an implementation ofcomplementary information, can improve capability of target recognition andanti-interference.Firstly, fusion algorithm for laser active and passive images was given. The imagefusion algorithm based on wavelet transform was given according to the characteristicsof images and the goal of fusion. Weighted average algorithm was used in the lowfrequency part to compensate laser image information, and then differences of thecoefficient were subtracted to remove the image background. In the high frequency part,we selected absolute value, whichever was greater.Secondly, the image fusion strategy for the range-gated imaging lidar was given.Data fusion of laser active intensity image and passive image retains rich information ofpassive optical image, and removes the background outside of the target. Then, DataSynthesis of fusion image and range image provided three-dimensional image.Again, we simulated the image fusion experiments. The complex three-dimensionalmodel was established, and simulation laser active and optical passive images as thetargets of image fusion were generated. The fusion of active and passive images wasattempted. We discussed effects of decomposition levels for image fusion, and obtainedthe optimal decomposition level for the simulation image was2through evaluationmethods combined subjective and objective. The results showed that the informationentropy of the image was higher than laser active image using improved fusion algorithm,that illustrated image fusion could compensate laser active images information, withoutintroduction of background interferences from passive images, so the goal of fusion wasachieved. Feasibility and advantages of fusion strategy and algorithm were validated bycomparison with other fusion algorithms.Finally, we conducted experimental validation of range-gated laser images fusion.We obtained the optimal decomposition level for the two target images that440and700meters away from imaging equipment were2and3, and further validated feasibility and advantages of the fusion strategy and algorithm through comparison of fused image andsource images by information entropy, average gradient, standard deviation, and otherevaluation factors. The results showed that image fusion could retain distanceinformation, and improve the clarity of distance image, and rich image information at thesame time.