Study On Auto-focus Of Industrial Camera And Image Mosaic And Their Applications To Robots For Bridge Health Detection

As the rapid development of bridges, the healthy detection of them grows to be more andmore important, among which the video detection brings a series of problems. This paperfocuses on study on Auto-focus of industrial camera and image mosaic and its application onbridge-detecting mobile robots. Using evaluation of image clarity based on ridge detectionand improved mountain climb servo(MCS), this paper achieved auto-focus technology. Basedon corner features, this paper achieved image auto-mosaic technology, and finally achievedthe application of the two technologies on the mobile robots. The main content includes:Evaluation method of image clarity based on improved Sobel operator: Analysis showsthat evaluation of image clarity based on image edge takes advantage of the feature that thegray gradient changes dramatically. This paper simulated kinds of traditional function used toevaluate image clarity, and concluded that the traditional evaluation functions are useless tothe noisy images. So this paper proposed a clarity evaluation method based on mean filter anda presetting threshold. The robustness to noise of this method is much better.The optimal focus position search method based improved MCS: This paper studiedtraditional pole search method and improved traditional MCS, finally found a efficient polesearch method.The corner detection method based on improved Harris operator: This paper analysed thekey elements in image mosaic, such as feature detection method, similarity measure methodand the method to remove wrong matching. In addition, this paper simulated kinds oftraditional corner detection method, and concludes that there are several problems in some oftraditional corner detection method,such as inaccurate positioning, bad robustness, and so on.This paper improved traditional Harris operator,and improved the robustness of cornerdetection method.This paper achieved corner matching based on mutual max correlation coefficientalgorithm and RANSAC algorithm, and finally achieved image registration. This paperanalysed image interpolation method and image fusion method, and achieve image seamlessmosaic based on mean value interpolation and fade in-fade out fusion.This paper achieved software and hardware implementation of the industrial camera’sauto-focus and image auto mosaic on the mobile robots, based on the mobile robot platformused to detect bridges.