Research On Digital Focusing In Multispectral Imaging Systems

Multispectral imaging system is com posed of multiple optical pass filter consisting of optical channel and imaging system, through to the reflectivity of each channel in turn to co llect, on behalf of the color information of the object image. So the multispectral image containing a large number of spectral energy which is applied to the image to the reconstruction of the reflectivity.In multispectral imaging system in use pr ocess, found that the system of a bug: filter in the system caused by the im aging system of optical path of change and the refractive index of the dif ferent filter error caused by the differences of i maging, affect the imaging quality is not good for the reflectivity of reconstruction image. So we need to f ind a m ultispectral image autofocus algorithm. This pape r selects the multispectral liquid crystal tunable filter(LCTFs) for image acquisition and combined with CMOS im aging sensor, due to the com pact design, low power consum ption, LCTFs can continuously quickly gathers the image of each band relative to general multispectral imaging system, the image quality will be higher, more suitable for our research.Definition of the ev aluation is the pr econdition of autofocus algorithm. Clarity evaluation function which can be roughly di vided into several categories: gray gradient function, inform ation function, fr equency domain function and statistical functions, etc. The gray gradient function is widely used, and the algorithm is more mature. Considering the evaluation of multi spectral images, using the genera l gray gradient function, such as gray scale dif ference algorithm, energy gradient algorithm, Tenengrad algorithm, etc., the ef fect is not good. According to the ref erence, this paper choose the m ost similar to the hum an eye subjective evaluation of the gray difference distribution as the image definition evaluation function.There are two kinds of auto focusi ng technology based on i mage processing. The first is the focus depth m ethod, the second is the depth of the depth of focus. Fuzzy multi spectral images are mainly caused by dispersion, for a band of the image, namely imaging system of di ffusion function makes the pi xel corresponding to different radius circle of confusion, considering the factors, the depth from defocus method is more suitable in the choice of our multi spectral imaging system. Because common sharpness enhancement filter is based on the image gray value of operations, will result in a gr eat deal of energy distribution of changes to the processing of the image. Therefore, this paper in th e premise of ensuring the m ulti spectral image energy distribution is essentially th e same, proceed from the frequ ency domain algorithm, focusing filter based on freque ncy domain Gaussian function of i mage enhancement. Taking into account the energy of the image is basically low frequency component, the details of the i mage information for the high frequency com ponent which is the fact that the design of the filter to maintain the low frequency constant amplification of the high-frequency component. Through the experim ental comparison of the para meters of the f ilter, the filter can achieve the best f iltering effect. Finally, by comparing the histogram of the image before and after filtering, to analyze whether the energy distribution of the image to achieve the basic premise of this premise.In the end of this article summarizes the research work and innovation of this thesis mainly points, and the problems and deficiencies and prospect the next task is proposed...