Entropy method for selection of characteristic parameters of optical filter defects

Abstract

The invention discloses an entropy method for selecting filter defect feature parameters, which includes: segmenting a circumscribed rectangle containing the defect from a defect filter image to form a defect ROI; setting the elements of the candidate feature set F, and setting the selected feature Set S is an empty set; calculate defect ROI feature value, construct sample set; calculate candidate feature f i f k The normalized mutual information SU(f i f k ,C); according to SU(f i f k , C) The maximum value selects the first element s of S 1 ;Remove the features that have been selected into S in the candidate feature set F and the normalized mutual information SU(f i f k , C) Candidate features less than the threshold; calculate each candidate feature f in the candidate feature set F i f k The evaluation function J(f i f k ,C,S) values; according to the evaluation function J(f i f k , C, S) maximum value to select the next element of the selected feature set S; remove the feature and evaluation function J(f i f k ,C,S) Candidate features less than the threshold; judge whether the candidate feature set F is an empty set; output the selected features.

Description

technical field [0001] The invention relates to the technical field of visual recognition of optical filter defects, in particular to a method for selecting characteristic parameters of optical filter defects using the principle of entropy. Background technique [0002] Optical thin film filters are widely used in optical communications, laser technology, optical imaging and detection, and play an important role in micro cameras, biomedical instruments, and advanced laser systems. For example, in the field of optical communication, optical filters are not only the key components for multiplexing and demultiplexing in wavelength division multiplexing systems, but also commonly used in gain flattening of erbium-doped fiber amplifiers, all-optical up and down channels, and wavelength switches. In optoelectronic products, each mobile phone camera must be equipped with a filter. There is a huge demand for optical filters in the domestic and international markets, and the domesti...

AI Technical Summary

Problems solved by technology

[0008] In order to solve the above-mentioned technical problems, the object of the present invention is to provide a kind of entropy method of filter defect feature parameter selection, this method considers the situation that some features have poor distinguishing ability when acting alone and strong distinguishing ability when combined with other features, According to the correlation between candidate features and defect categories, and the amount of new information provided by candidate features for defect classification, an evaluation function for judging the importance of candidate features is designed. Selection of Characteristic Parameters of Optical Sheet Defects

Images