Improved image enhancement method based on block matching and recovery and combined trilateral steering filtering

A technology of guided filtering and image enhancement, applied in the field of image processing, can solve the problems of poor visual effect, poor preservation of structural information, loss of structural information, etc., to achieve the effect of removing pseudo-texture and noise, and achieving good visual effect.

Inactive Publication Date: 2017-03-08
XIDIAN UNIV
View PDF2 Cites 27 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This method improves the stability and accuracy of depth image inpainting, but because the algorithm needs to create a 3600×3600 matrix for each 60×60 pixel block, the time and space complexity are high
[0005] Xiaojin Gong of Zhejiang University, etc. published "Guided depth enhancement via a fast marching method" (GFMM) on ELSEVISER's Image and Vision Computing in 2013, and proposed a method based on the Fast Marching Method (FMM for short) proposed by A.Telea in 2004. Guided depth enhancement algorithm, which is based on the original FMM algorithm, introduces the corresponding color image to optimize the expansion mechanism. Difference
In addition, Junyi Liu et al. published "GuidedDepth enhancement via Anisotropic Diffusion" (GAD) on Springer International Publishing Switzerland in 2013, which transformed the depth image enhancement problem into a linear anisotropic diffusion problem, and used a sparse linear system to solve the image enhancement problem. Compared with GFMM, this method can better preserve the structural information, but there are a lot of noise and pseudo-texture in the resulting image, and the visual effect is poor.
[0006] To sum up, the existing deep image enhancement technology has problems such as serious loss of structural information, high image distortion, high image processing time and space complexity, and cannot meet the needs of practical applications.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Improved image enhancement method based on block matching and recovery and combined trilateral steering filtering
  • Improved image enhancement method based on block matching and recovery and combined trilateral steering filtering
  • Improved image enhancement method based on block matching and recovery and combined trilateral steering filtering

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] see figure 1 , the improved block matching repair of the present invention and joint trilateral guided filter image enhancement method, the enhancement process includes the following steps:

[0041] (1) Input image: input the original depth image collected and save it as a grayscale image, see image 3 . According to the Kinect image acquisition mechanism, the region formed by pixels with invalid pixel values ​​is an unknown region, and the remaining pixels, that is, the region formed by valid pixel values, is a known region. In this embodiment, the depth image is saved as 8 bits Grayscale image, so the pixel value of the invalid pixel is 0, and the area formed by the pixels whose pixel value is not 0 is the known area.

[0042] (2) Original image preprocessing: For the collected original depth image, use the morphological closing operation to perform preprocessing to remove the random depth missing points existing in the original image; see image 3 , the pixels wit...

Embodiment 2

[0053] Improved block matching repair and joint trilateral guided filtering image enhancement method are the same as embodiment 1, wherein the block matching priority estimation function P described in step (4):

[0054] The block matching priority estimation function P(p) of each pixel point p is defined as follows:

[0055] P(p)=C(p)D(p)L(p) (1)

[0056] Among them, C(p) represents the confidence of point p, that is, the proportion of known pixels in the neighborhood centered on p. The larger the ratio, the greater the number of known pixels in the pixel block centered on p. , that is, the more known information used to predict the pixel value of an unknown pixel, the more accurate the prediction result; D(p) represents the data item of point p, ensuring that the block close to the normal direction is repaired earlier; L(p) represents The level set distance factor is defined by the diffusion time function, thus ensuring that the closer the pixel is to the boundary of the un...

Embodiment 3

[0071] Improved block matching repair and joint trilateral guided filtering image enhancement method are the same as embodiment 1-2, wherein the joint trilateral guided filtering method described in step (9):

[0072] The filtering model of the joint trilateral guided filtering method is defined as follows:

[0073]

[0074] Among them, ω p As a normalization factor, the weighted sum of each factor is guaranteed to be 1, defined as shown in formula (8); JiontTF[I] pIndicates the result obtained by filtering the pixel block I centered on p by using joint trilateral guided filtering, I is the pixel block to be repaired, and I q is the intensity value of pixel point q, s is the neighborhood centered on p; parameter σ s is the size of the Gaussian kernel in the spatial domain, Computes the Gaussian weights in the spatial domain, which, like ordinary Gaussian filters, decrease as the spatial distance between p and q increases; I cp and I cq are the pixel values ​​of pixel ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an improved image enhancement method based on block matching and recovery and combined trilateral steering filtering. With the method, removing of lots of holes and severe noises in a collected depth image is realized. The method comprises: pretreatment of an original image is carried out; a to-be-recovered unknown region in the image after pretreatment is marked; a pixel point priority level is calculated; a to-be-recovered block is selected; an optimal matching block is searched; recovering is carried out; determination is carried out; the image is processed by using combined trilateral steering filtering; and image enhancement processing is ended and a result is outputted. According to the method provided by the invention, with introduction of a horizontal set distance factor and a diffusion time function, a block-matching-based image recovery sequence is improved and a structural hole is filled; and on the basis of the combined trilateral steering filtering, a pseudo texture generated by recovering and lots of original noises are removed. In terms of a visual effect and a quantitative analysis, the texture structure of the original image can be kept well for the image after enhancement processing and noise information in the image is removed, so that the image distortion degree is low.

Description

technical field [0001] The invention belongs to the technical field of image processing, mainly relates to depth image enhancement, and specifically provides an improved image enhancement method of block matching repair and joint trilateral guide filtering, which can be used for hole filling and noise removal of depth images collected by Kinect and the like. Background technique [0002] In recent years, the cheapness of depth image acquisition equipment has provided the possibility for the acquisition and use of depth information, and also brought opportunities for the further development of applications involving depth information in robotics, especially in map construction, path planning, and environmental perception. etc., rely heavily on the effective use of depth information. [0003] However, as one of the inexpensive depth information collection devices, Kinect mainly relies on parallax images and triangulation principles to obtain depth information in the scene, so ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06T5/00
CPCG06T5/005G06T5/002G06T2207/20036
Inventor 宋娟张亮张淑娥沈沛意朱光明
Owner XIDIAN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap