Denoising method of low illumination underwater motion image based on improved canny
Introduction
At present, complex domain images are widely used in measurement, such as in the field of aerospace, terrain elevation acquisition, terrain subsidence analysis, etc. Through the complex data of synthetic aperture radar interferometry, and in the field of medicine, pathological analysis can be carried out through the complex MRI images [1]. However, whether SAR or MRI images, there are a lot of noise in the measurement process due to the external environment, the limitations of measurement equipment and other factors, therefore, it is of great significance to study the denoising algorithm of complex image [2,3].
In reference [4], a complex image denoising algorithm based on group sparse coding is proposed. Complex values are grouped and sparse coded as a unified whole, and image blocks in the same group are limited to use similar elements in the training dictionary. The algorithm is verified by simulating the real InSAR and MRI images. The experimental results shows that the algorithm can get lower noise reduction error, Especially for the image with large smooth area or the image with high noise level, it has a great advantage in noise reduction, but it has the disadvantage of low SNR. In reference [5], an image denoising algorithm based on dual tree CWT and adaptive bilateral filter is proposed. The algorithm uses dual tree complex wavelet transform to decompose the noisy image in multi-scale and multi-directional, the high-frequency coefficients of each direction subband are quantized by the improved threshold, and the low-frequency subband coefficients of each scale are filtered by the adaptive bilateral filter, the reconstructed image is further denoised. The simulation results show that this method has good filtering effect on the mixed noise and can protect the edge details of the image, but there is a problem of low similarity of the image structure after noise reduction. In reference [6], a dictionary algorithm based sparse speckle denoising method for OCT image is proposed. Firstly, the algorithm performs logarithmic transformation for OCT image, sparse coding with orthogonal matching tracking algorithm, and adaptive dictionary updating with K SVD learning algorithm. Finally, it returns to airspace through weighted average and exponential transformation. The experimental results show that the algorithm can effectively reduce the speckle noise in OCT image and obtain good visual effect, but there is also the problem of low similarity of image structure after noise reduction.
In order to make better use of these low illumination underwater moving images, it is inevitable to use image enhancement and other technologies to deal with these problems. Image enhancement technology is an important information of purposeful reproduction of image, and it is also the pre-processing stage of image. It is mainly to enhance the recognition effect of image and improve the characteristics of interest in image. The main purpose of low illuminance underwater motion image enhancement is to make the enhanced image more in line with people's subjective visual experience, and the image can be easily analyzed and processed by computer vision and other equipment. In the enhancement of low illuminance underwater moving image, the most important thing is to reduce the noise, so as to enrich the details of the image, and make the enhancement results more in line with the needs of human senses or the analysis of computer vision processing system. Therefore, a denoising method of low illuminance underwater motion image based on improved canny is proposed. Canny edge detection operator is a multilevel edge detection algorithm developed by John F. canny in 1986, It can significantly reduce the data size of the image while retaining the original image attributes, and has a wide range of applications in image noise reduction. According to the results, it is proved that the method has a high signal-to-noise ratio, and the highest value of the structural similarity of the noise reduction method is 0.92, which proves that the structure similarity of the image after noise reduction and the original image is high, and the visual effect of noise reduction is good, which fully verifies that the method has a good visual effect Practical application of image denoising.
Section snippets
Target rough segmentation based on region and edge feature fusion
The underwater moving image is affected by the scattering and absorption of the light by the water medium, the change of the light conditions, the imaging distance and the hovering and shaking of the auvms dynamic positioning. The underwater moving image often has the characteristics of low illumination, such as high degree of blur and serious noise. In addition, because the underwater camera imaging is interfered by other equipment in the underwater acquisition system, the underwater moving
Extraction of local intuitionistic fuzzy entropy
Based on the result of image segmentation, local intuitionistic fuzzy entropy is extracted on the basis of intuitionistic fuzzy ife, and it is introduced into anisotropic diffusion model to realize image denoising. The low illumination underwater moving image is regarded as a fuzzy set, and each element is composed of membership value, non membership value and hesitation value. Therefore, the image with size and gray level is described as fuzzy set :
Experimental research and analysis
In order to verify the effectiveness of the low illumination underwater motion image denoising method based on the improved canny, a comparative experiment of low illumination underwater motion image denoising is carried out by using the methods of reference [4], reference [5] and reference [6], and PSNR (Peak Signal-to-Noise Ratio) and SSIM (Structural SIMilarity) are used to evaluate the experimental results.
Conclusion
In order to improve the efficiency of image denoising, this paper proposes a method of image denoising based on improved canny. The region and edge feature fusion method is used to segment the target roughly, and then the fusion threshold of image channel is segmented. According to the result of image segmentation, local intuitionistic fuzzy entropy is extracted on the basis of intuitionistic fuzzy, which is introduced into anisotropic diffusion model and diffusion function. The partial
Declaration of Competing Interest
No conflict of interest.
Acknowledgement
The Research is Supported by Science and Technology Department of Shanxi Province - Experimental Study on Psychological Effects of Exercise Entervention for Compulsory Male Drug Abusers (No. 201803052).
Hui Lv, female, was born in May 1979, associate professor, Ph.D., and graduated from Shanxi University, majoring in Physical Education and Training. She works in College of Physical Education, Shanxi University now, and the research areas are School Physical Education and Early Childhood Sports. She published 19 papers and presided over 5 provincial projects.
Reference (24)
- et al.
Image denoising algorithm based on Dual-Tree CWT and adaptive bilateral filtering
Comput Eng Appl
(2018) - et al.
Rigel: flexible multi-rate image processing hardware
ACM Trans Graph
(2016) - et al.
Automatically scheduling halide image processing pipelines
ACM Trans Graph
(2016) - et al.
Halide: decoupling algorithms from schedules for high-performance image processing
Commun ACM
(2017) - et al.
A denoising method of the complex valued images based on grouped sparse coding
Chinese J Comput
(2019) - et al.
OCT image speckle sparse noise reduction based on dictionary algorithm
Opto-Elect Eng
(2019) - et al.
Monochromatic-beam-based dynamic X-ray microtomography based on OSEM-TV algorithm
J Xray Sci Technol
(2017) - et al.
Smartphone and mobile image processing for assisted living: health-monitoring apps powered by advanced mobile imaging algorithms
IEEE Signal Process Mag
(2016) - et al.
EMHP: an accurate automated hole masking algorithm for single-particle cryo-EM image processing
Bioinformatics
(2017) - et al.
Identification of low coronal sources of “stealth” coronal mass ejections using new image processing techniques
Astrophysical J
(2017)
Image processing using proper orthogonal and dynamic mode decompositions for the study of cavitation developing on a NACA0015 foil
Exp Fluids
A simple setup for episcopic microtomy and a digital image processing workflow to acquire high-quality volume data and 3D surface models of small vertebrates
Zoomorphology
Cited by (5)
Removal of Impulse Noise in Grayscale Images Under Fuzzy Knowledge Measure
2023, Tien Tzu Hsueh Pao/Acta Electronica SinicaVeracious Interpolated Measure of Angle and Length for Underwater Motion Blurred Images
2023, Fluctuation and Noise LettersApplication of Image Processing Based on Improved Canny Algorithm and Gurukin's Theorem
2023, ACM International Conference Proceeding SeriesUnsupervised Change Detection in Wide-Field Video Images Under Low Illumination
2023, IEEE Transactions on Circuits and Systems for Video TechnologyVideo Image Processing Based Intersection Vehicle Queue Length Detection
2022, Lecture Notes on Data Engineering and Communications Technologies
Hui Lv, female, was born in May 1979, associate professor, Ph.D., and graduated from Shanxi University, majoring in Physical Education and Training. She works in College of Physical Education, Shanxi University now, and the research areas are School Physical Education and Early Childhood Sports. She published 19 papers and presided over 5 provincial projects.
Haijun Li, male, was born in February1978, lecturer, master's degree, and graduated from Shanxi University, majoring in Physical Education and Training. He works in College of Physical Education, Taiyuan University of Technology now, and the research area is Football Training Theory and Methods. He published5 papers in General Publications and presided over a provincial project.