The Method Of Adaptive Illumination Compensation And Measurement For Target Object Based On The Projector-camera System

Illumination is one of the key factors determining the quality of machine vision.It is easy to produce both the underexposed region and the overexposed region on the same image when we collected complex scenes influenced by external light sources,the material of the object itself and the principle of refraction and refraction.The region cannot be restored by image processing technology.Because of the acquisition errors,the following operations such as image recognition,information retrieval,defect detection and others cannot be performed.For this situation,this paper proposes an adaptive adjustment of the illumination,the illumination intensity of the underexposed region is enhanced and the illumination intensity of the overexposed region is reduced.This theory is based on a projector with a controllable light source,a CCD camera capable of acquiring information,and a processor between the projector and the camera.In this system,the projector obtains adaptive projection light through iterations,and the camera collects scenes with good illumination and clear display,laying the foundation for the next operation.In this paper,we first use the Fourier fringe analysis method in 3D shape measurement to make the projector-camera coordinate system one-to-one for each pixel The advantages and disadvantages of various Fourier fringe analysis methods are analyzed,and a technique more suitable for this subject is found,namely a random two-step phase shift fringe analysis method.This method can greatly improve the accuracy under the premise of slightly increasing the complexity.Secondly,the previous adaptive iterative algorithm are discussed and verified.Based on the Inversion Iterative Method and the Filter Iteration Method,the new iterative techniques--the Difference Iterative Method and the Saliency Detection Iterative Method are proposed.These two methods make up for the lack of image quality evaluation techniques,the number of iterations,the inability to automatically stop iterations,and the need for manual assistance of traditional iterative algorithms.And through the Improved Absolute Central Moment Algorithm and the Non-reference Image Quality Evaluation Method BRISQUE,the image quality is evaluated,and the whole system is fully automated from input to output.Finally,the validity of the new iterative algorithms is verified by theoretical analysis and experiments.The number of iterations is reduced from 40 times to 20 times and less than 10 times,which greatly reduces the computation time and proves the validity and innovation of this subject.