All - Digital Focusing Method And Evaluation Of Dispersion Multispectral Imaging System

The spectrum of body surface reflection decide the object color,different from the traditional digital camera,multi-spectral imaging system is composed of multiple optical pass filter to make optical channel,each channel gathering the reflectivity on behalf of the object and then the color of the image is contained.Compared with the general color digital images,it cantains more spectral information,naturally in favor of the reconstruction of the image reflectance.But no matter how accurate of the multi-spectral imaging system,they all have common shortcomings,filter will change the optical pass,the thickness of filter and different reflectivity also contribute to imaging errors,leaing to the image blur,which will affect the measurement of its application in filed.Based on above various reasons,it is particularly important to propose an autofocus algorithm for multi-spectral image fuzzy.This paper choose multi-spectral liquid crystal tunable filter(LCTFs) for image image acquisition and combined with CMOS sensor imaging,due to the compact design,low power consumption,LCTFs can gathers the images continuously and quickly in each band,the imaging quality is higher than general multi-spectral system,so it is appropriate from the perspective of the imaging processing idea to realize the multi-spectral autofocus.The autofocus technology can based on imaging processing can be roughly devided into two kinds,as depth of focus and away from the focal depth.The fuzzy multi-spectral images acquised by LCTFs are basically due to dispersion caused by the factors.For a certain band image,also called the point spread function of imaging system makes the pixels corresponding to fuzzy circle radius range,the situation is suitable for the principle of away from the focal depth method to design filter,away from the focal depth,the greater the fuzzy circle radius is,then the sharpness of images decreases.The article from the perspective of frequency domain to design image enhancement focus filter.The premise is to ensure that the energy distribution of multi-spectral images no change,because of the principle that the energy of the images almost in the low frequency component and the details of the images almost in high frequency component for the modulation of focal filter we designed.From the gray histogram of before and after processing image contrast to find the most suitable parameters,which is the autofocus image we need.Sharpness problem is to determine whether a multi-spectral image of the band get accurate focusing premise,for the image sharpness evaluation basically based on subjective judgement of human visual system,but the channel and the quantity of multi-spectral images is large,coupled with many uncontrolled factors of subjective evaluation,so we compared several kinds of objective evaluation algorithm to simulate human visual system, the experimental data shows that gray difference function is can simulate hunman visual system and also has advantages such as fast efficient.Finally,use this objective evaluation algorithm to calculate sharpness value before and after treatment then draw the conclusion.In the end of this article,summarize the mainly research of thesis and the innovation points,find the problem and the deficiencies,put forward the next task and prospect.