Noisy image feature extraction method

• Main Authors: WANG MINGDA, WANG WENYAO, ZHU YOUCHAN
• Format: Patent
• Language: Chinese; English
• Published: 21.04.2020
• Subjects: CALCULATING; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention provides a method for extracting image features after denoising of a noisy image. The method comprises the steps of determining a to-be-processed target image; based on a denoising auto-encoding (DAE) algorithm principle, denoising the noisy image, and adding an image enhancement effect as an image feature extraction sample; extracting image local features based on an improved accelerated robust feature SURF extraction algorithm; and taking the extracted features as image features of the image. Application of embodiments of the present invention, a method of denoising a noisy image based on a DAE algorithm and adding an image enhancement operation is combined with an improved SURF algorithm; the method is advantaged in that the image sample characteristic effect is enhanced,the dual denoising effect is realized, the characteristic point detection method is optimized to extract the target image characteristics, the operation rate is improved, the characteristic extractioncalculation amount is red