公开(公告)号：US20180053287A1

公开(公告)日：2018-02-22

申请号：US15305350

申请日：2015-11-09

Applicant: China University of Mining and Technology

Inventor： Deqiang CHENG ,  Hai LIU ,  Leida LI ,  Guopeng ZHANG ,  Xingge GUO

IPC: G06T5/00 ,  G06T5/20

CPC classification number: G06T5/008 ,  G06T5/002 ,  G06T5/20 ,  G06T2207/10016 ,  G06T2207/10024 ,  G06T2207/20182 ,  G06T2207/20201 ,  G06T2207/20208

Abstract: The invention relates to image processing technology field, and discloses a video image denoising and enhancing method based on random spray retinex, including: structuring spray pixel sets, and tuning parameters related to the random spray retinex based on the spray pixel sets, wherein the parameters include quantity of the spray pixel sets and quantity of pixels; processing video images with random spray retinex based on tuned parameters; denoising the video images processed by the random spray retinex via low pass filters and blur channels to get a brightness variation calculating formula; obtaining a brightness calculating, formula of output images, combined with the brightness variation calculating formula, and calculating brightness variations of three channels via the brightness calculating formula to get local brightness estimating vectors; and fusing the three channels based on the local brightness estimating vectors to get denoised and enhanced video images.

公开(公告)号：US20180365843A1

公开(公告)日：2018-12-20

申请号：US15739779

申请日：2015-11-09

Applicant: CHINA UNIVERSITY OF MINING AND TECHNOLOGY

Inventor： Deqiang CHENG ,  Leida LI ,  Hai LIU ,  Guopeng ZHANG ,  Wei CHEN ,  Songyong LIU

IPC: G06T7/246 ,  G06T7/13 ,  G06K9/00 ,  G06K9/80 ,  G06K9/62

CPC classification number: G06T7/248 ,  G06K9/00718 ,  G06K9/6202 ,  G06K9/80 ,  G06K2209/21 ,  G06T7/13 ,  G06T7/20 ,  G06T2207/10016 ,  G06T2207/20221

Abstract: A method for tracking a moving target based on an optical flow method, including, providing video images, and implementing pre-processing of the images to generate pre-processed images; implementing edge-detection of the pre-processed images and using an optical flow method to extract target information from the pre-processed images, and on the basis of a combination of the edge-detection information and the extracted target information, generating a complete moving target; using an optical flow method to perform estimation analysis of the moving target and using a forward-backward error algorithm based on feature point trace to eliminate light-generated false matching points; and creating a template image and implementing template image matching to track the moving target. The method and system for tracking a moving target based on an optical flow method have the advantages of accurate and complete extraction and the ability to implement stable tracking over a long period of time.