CN114708180A - Bit depth quantization and enhancement method for pre-distorted image with dynamic range preservation - Google Patents
Bit depth quantization and enhancement method for pre-distorted image with dynamic range preservation Download PDFInfo
- Publication number
- CN114708180A CN114708180A CN202210398259.0A CN202210398259A CN114708180A CN 114708180 A CN114708180 A CN 114708180A CN 202210398259 A CN202210398259 A CN 202210398259A CN 114708180 A CN114708180 A CN 114708180A
- Authority
- CN
- China
- Prior art keywords
- image
- bit
- quantization
- bit image
- dynamic range
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000013139 quantization Methods 0.000 title claims abstract description 55
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000004321 preservation Methods 0.000 title claims abstract description 10
- 238000012545 processing Methods 0.000 claims abstract description 16
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 6
- 238000012423 maintenance Methods 0.000 claims abstract description 5
- 238000010606 normalization Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 3
- 230000014759 maintenance of location Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 241001270131 Agaricus moelleri Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20132—Image cropping
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a bit depth quantization and enhancement method for a pre-distorted image with dynamic range maintenance, and belongs to the technical field of image processing. Firstly, normalizing an input high-bit image, then calculating a predistortion template image and adjusting the resolution of the predistortion template image to obtain a predistortion image, then adding the normalized high-bit image and the predistortion image pixel by pixel, then calculating a quantization function with a dynamic range retention characteristic according to the bit depth of the low-bit image to quantize the predistortion high-bit image, finally taking the low-bit image as an input image of a network, and performing bit depth enhancement on the low-bit image through a convolutional neural network to obtain the high-bit image. The quantization method with dynamic range preservation can inhibit the blurring of over-bright or over-dark areas in the reconstructed image and improve the image reconstruction quality.
Description
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to a bit depth enhancement method of a pre-distorted image with dynamic range maintenance.
Background
As a modern mainstream information carrier, images bring convenience to daily life and also bring great challenges to storage and transmission equipment. In addition, in some application scenarios like wireless sensor networks with asymmetric system complexity, the computational resources at the data receiving end are not limited, and the computational resources at the data transmitting end are limited, so that it is necessary to reduce the data volume at the image transmitting end. By quantizing a high-bit image (i.e. a high-bit depth image, the bit depth of the image is a gray scale range that can be expressed by each channel of a pixel, for example, an image with the bit depth of n can represent 2n gray scales, the color representation is more delicate, and the visual experience of the image is better) into a low-bit image, simple compression of the image can be realized, however, reducing the bit depth of the image can bring about the problems of color distortion and false contour, which will seriously affect the viewing experience. Increasing the quantization step size can improve the compression efficiency, but introduces more severe false contours and color distortion, so that even the best-performing neural network method cannot guarantee effective suppression of the two artifacts. For this reason, the joint reconstruction method obtains a low-bit image with less artifacts by applying predistortion at the time of low-bit image generation, which is advantageous for reconstructing a high-quality high-bit image. However, in the existing quantization method, due to the loss of the dynamic range, the boundary of the over-bright area and the over-dark area of the high-bit image generated by the depth learning-based joint reconstruction method is obviously blurred, and the subjective visual experience is influenced.
Disclosure of Invention
The invention aims to provide a bit depth quantization method with dynamic range maintenance, which can further improve the bit depth enhancement performance, realize the suppression of obvious blurring of a reconstructed high-bit image in an over-bright area and an over-dark area, and enable the reconstructed high-bit image to achieve the best subjective and objective quality.
The technical scheme adopted by the invention is as follows:
a bit depth quantization and enhancement method for a pre-distorted image with dynamic range preservation, comprising the steps of:
step S1: and (3) carrying out quantization processing on the high-bit image:
step S101: carrying out normalization processing on an input high-bit image: dividing the gray value of each pixel by the maximum gray value of the high bit image;
the quantization step length Q is 1/(2) calculated based on the bit depth l of the quantization targetl-1), namely, the high bit image is quantized to obtain a low bit image with bit depth of l;
step S102: calculating to obtain a pre-Distortion function (PSD) template image according to the quantization step length Q;
step S102: performing non-overlapping tiling processing on the obtained template image to obtain a pre-distorted image with the image resolution consistent with the high-bit image;
step S103: adding the pre-distortion image and the high-bit image subjected to normalization processing pixel by pixel to obtain a pre-distortion high-bit image;
step S104: setting a quantization mode with dynamic range maintenance according to the quantization step length Q, and quantizing the pre-distorted high-bit image to obtain a low-bit image;
step S2: establishing a mapping relation from a low-bit image to a high-bit image through a convolutional neural network to obtain an image bit depth enhancement network; the low-bit image obtained in step S1 is used as an input to the image bit depth enhancement network, and an enhanced image, i.e., a reconstructed high-bit image, is obtained based on the output.
Further, in step S104, the quantization method with dynamic range preservation specifically includes:
if the gray value of the current pixel is less than or equal to Q/2, the gray value of the current pixel is quantized to 0;
if the gray value of the current pixel is greater than 1-Q/2, the gray value of the current pixel is quantized to 1;
and if the gray value of the current pixel is in [ Q/2,1-Q/2], uniformly dividing the value range [ Q/2,1-Q/2] into a plurality of subintervals, and obtaining the quantization value of the current subinterval based on the midpoint of each subinterval.
Further, in the step S104, the value range is [ Q/2,1-Q/2]Is the new quantization step Δ ═ 1-Q)/(2 at quantizationl-2) each ofThe quantization value of the subinterval is [ Q + (2k +1) Delta](iii)/2, wherein k is 0,1, … 2l-3。
Further, in S102, the template image is 3 × 3, and in any 3 × 3 neighborhood, is a complete set of { Q (1-L)/2L, Q (3-L)/2L, …, Q (L-1)/2L }, where L ═ 2n +1)2，n＝1。
That is, the quantization method with dynamic range preservation adopted in the present invention belongs to a non-uniform quantization method, and the gray scale value range of the high-bit image processed by the pre-distortion template image is less than 0 and greater than 1, which exceed the specified range [0,1 ] respectively]Outer Q/2, so that the new value range is modified to [ -Q/2,1+ Q/2 [ -Q/2 [)]. When l bit quantization is performed on the image, the total number is 2lA quantization section in which a gradation value is quantized to 0 when the pre-distorted high-bit image is in a range of Q/2 or less, to 1 when the pre-distorted high-bit image is in a gradation range of more than 1-Q/2, and to a middle range [ Q/2,1-Q/2] when the pre-distorted high-bit image is in the middle range]Then, the range is first uniformly quantized to obtain a new quantization step Δ ═ 1-Q)/(2n-2), each interval taking its midpoint as the quantized value, i.e. [ Q + (2k +1) Δ]/2. The pre-distorted high bit image is quantized in this way to obtain a low bit image, which can be expressed as a function:
wherein, ILRepresenting a quantized low bit-rate image, IPSDRepresenting a pre-distorted high bit image.
The technical scheme provided by the invention at least has the following beneficial effects:
the improved quantization method adopts non-uniform quantization to the brightest area and the darkest area and adopts uniform quantization to the middle area, thus achieving the purpose of maintaining the dynamic range, and further inhibiting the blurring problem of the reconstructed high-bit image in the over-bright area and the over-dark area.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a schematic processing diagram of a bit depth quantization and enhancement method for a pre-distorted image with dynamic range preservation according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a pre-distorted template image in an embodiment of the invention;
fig. 3 is a diagram illustrating a quantization process in an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
The bit depth enhancement algorithm based on the PSD belongs to a joint reconstruction algorithm, a generation process of a low-bit image and a reconstruction process of a high-bit image are considered at the same time, the image is subjected to pre-distortion treatment before quantization to construct effective multi-observation of signals, the bit depth is favorably reconstructed by a convolutional neural network in a subsequent stage, and the image reconstruction quality is improved. When the method is applied to image transmission processing, high-bit images can be compressed through quantization processing provided by the embodiment of the invention, then image transmission processing is carried out, and low-bit to high-bit image enhancement is carried out on a receiving end based on the enhancement mode provided by the embodiment of the invention, so that high-bit images with higher quality are reconstructed on the receiving end, and display conversion processing of the received low-bit images in a high-bit display is realized.
As shown in fig. 1, a specific implementation process of the bit depth quantization and enhancement method for a pre-distorted image with dynamic range preservation provided by the embodiment of the present invention includes: the method comprises the steps of normalization processing of a high-bit image, calculation of a pre-distortion function template image (pre-distortion template image), acquisition of the pre-distortion image, calculation of a quantization function, quantization of the pre-distortion image and bit depth reconstruction based on a neural network.
Step 1: for high bit image IHI.e. an image having a first image data format, according to a low bit image (an image having a second image data format with an image bit depth lower than the first image data format) ILIn this embodiment, assuming that the bit depth of the high-bit image is 16 and the bit depth of the low-bit image is 2, the maximum gray value that can be represented by the 16-bit depth is 216-1, after normalization, is l'H＝IH/(216-1), quantization step size Q1/22＝0.25；
Step 2: a pre-distorted template image is calculated from the quantization step Q0.25, which is the complete set of { -4/9Q, -3/9Q, -2/9Q, -1/9Q,0,1/9Q,2/9Q,3/9Q,4/9Q } in any 3 × 3 neighborhood, i.e. the 9 pixel values are randomly arranged in the 3 × 3 template. As a preferred mode, the method is arranged as shown in fig. 2, that is, the gray value of the central point of the pre-distortion template image is 0, the gray values of the adjacent pixel points of the central point are distributed in central symmetry with respect to the central point, and the values of the two symmetrical pixel points are opposite;
and step 3: assuming that the resolution of the high-bit image is M × N, before the high-bit image is quantized into the low-bit image, the pre-distorted template image calculated in step 2 is tiled from left to right and from top to bottom without overlapping and omission to obtain a pre-distorted image with the resolution of M × N, and the pre-distorted image is set to be δpsdIt is compared with a high bit image IHA pixel-by-pixel addition, denoted as I, is performedPSD＝IH+δpsd；
As a possible implementation manner, in this step, when obtaining the pre-distorted image, the pre-distorted template image may be first tiled from left to right and from top to bottom without overlapping and omission, where the size of the tiled image is greater than or equal to mxn, and if the size of the tiled image exceeds mxn, the redundant portion is cut off, so as to obtain an mxn pre-distorted image.
As a possible implementation manner, an mxn pre-distortion image may also be obtained by using a sliding window manner, where the size of a sliding window is the same as that of the pre-distortion template image, the step length is the side length of the pre-distortion template image, pixels of the pre-distortion template image are filled pixel by pixel every time the sliding window slides once in the mxn image range, and when the sliding window slides to the extreme end each time, the window beyond the image range is skipped, that is, no pixel filling is performed, so as to obtain the mxn pre-distortion image.
And 4, step 4: the quantization function with dynamic range retention is designed according to the quantization step Q of 0.25 as follows:
wherein, ILRepresenting the quantized image. The method is a non-uniform quantization method, different quantization modes are adopted for two end and middle regions, and the gray value range of a high-bit image processed by a pre-distortion template image is less than 0 and greater than 1, and exceeds a specified range [0,1 ] respectively]Outer 0.125, so the new value range is modified to [ -0.125,1.125]. 2bit quantization divides the whole interval into 4 segments, and quantizes the predistortion high bit image in the way to obtain a low bit image ILThe quantization diagram is shown in fig. 3.
And 5: at an image receiving end, a nonlinear mapping relation from a 2-bit predistortion low-bit image to a 16-bit high-bit image is established through a convolution neural network.
The convolutional neural network may adopt any conventional network structure in the field, and the embodiment of the present invention is not particularly limited, for example, a simple and efficient neural network such as EBDA-CNN is selected, and a high-bit image with good host and guest quality is reconstructed from a low-bit image, wherein the evaluation mode for the image quality mainly includes a peak signal-to-noise ratio (PSNR) and a Structural Similarity Index (SSIM), and a calculation formula is shown as follows.
Wherein, IHRepresenting the original 16-bit image of the picture,
representing a 16bit image reconstructed by the network. The larger the PSNR, the better the quality of the bit-depth reconstructed image, and vice versa.
Wherein, muxRepresenting an original high bit image IHMean value of (d) (. mu.)yRepresenting bit-depth reconstructed high-bit images
The average value of (a) of (b),
and
representing the variance, σ, between the original image and the restored image, respectivelyxyRepresents IHAnd
covariance of (C)1And C2Is a constant. According to a formula, if the difference between a high-bit image obtained by a bit depth reconstruction algorithm and an original high-bit image is smaller, the quality of an output image is better, and the algorithm effect is better.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
What has been described above are merely some embodiments of the present invention. It will be apparent to those skilled in the art that various changes and modifications can be made without departing from the inventive concept thereof, and these changes and modifications can be made without departing from the spirit and scope of the invention.
Claims (4)
1. A bit depth quantization and enhancement method for a pre-distorted image with dynamic range preservation, comprising the steps of:
step S1: and (3) carrying out quantization processing on the high-bit image:
step S101: carrying out normalization processing on an input high-bit image: dividing the gray value of each pixel by the maximum gray value of the high bit image;
the quantization step length Q is 1/(2) calculated based on the bit depth l of the quantization targetl-1)；
Step S102: calculating according to the quantization step length Q to obtain a pre-distortion function template image;
step S102: performing non-overlapping tiling processing on the obtained template image to obtain a pre-distorted image with the image resolution consistent with the high-bit image;
step S103: adding the pre-distortion image and the high-bit image subjected to normalization processing pixel by pixel to obtain a pre-distortion high-bit image;
step S104: setting a quantization mode with dynamic range maintenance according to the quantization step length Q, and quantizing the pre-distorted high-bit image to obtain a low-bit image;
step S2: establishing a mapping relation from a low-bit image to a high-bit image through a convolutional neural network to obtain an image enhancement network; the low bit rate image obtained in step S1 is used as an input to the image enhancement network, and an enhanced image is obtained based on the output.
2. The method according to claim 1, wherein in step S104, the quantization mode with dynamic range preservation is specifically:
if the gray value of the current pixel is less than or equal to Q/2, the gray value of the current pixel is quantized to 0;
if the gray value of the current pixel is greater than 1-Q/2, the gray value of the current pixel is quantized to 1;
and if the gray value of the current pixel is in [ Q/2,1-Q/2], uniformly dividing the value range [ Q/2,1-Q/2] into a plurality of subintervals, and obtaining the quantization value of the current subinterval based on the midpoint of each subinterval.
3. The method of claim 2, wherein in step S104, the value range is [ Q/2,1-Q/2]]Is the new quantization step Δ ═ 1-Q)/(2 at quantizationl-2), the quantization value of each subinterval being [ Q + (2k +1) Δ [ ]](iii)/2, wherein k is 0,1, … 2l-3。
4. The method according to any one of claims 1 to 3, wherein in step S102, the template image is 3 x 3 and in any 3 x 3 neighborhood is the complete set of { Q (1-L)/2L, Q (3-L)/2L, …, Q (L-1)/2L }, where L ═ 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210398259.0A CN114708180B (en) | 2022-04-15 | 2022-04-15 | Bit depth quantization and enhancement method for predistortion image with dynamic range preservation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210398259.0A CN114708180B (en) | 2022-04-15 | 2022-04-15 | Bit depth quantization and enhancement method for predistortion image with dynamic range preservation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114708180A true CN114708180A (en) | 2022-07-05 |
| CN114708180B CN114708180B (en) | 2023-05-30 |
Family
ID=82174881
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210398259.0A Active CN114708180B (en) | 2022-04-15 | 2022-04-15 | Bit depth quantization and enhancement method for predistortion image with dynamic range preservation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114708180B (en) |
Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101854714A (en) * | 2010-05-13 | 2010-10-06 | 清华大学 | Method for achieving wireless communication timing coarse synchronization by using 1bit quantification and hard decision |
| CN103295208A (en) * | 2013-05-09 | 2013-09-11 | 浙江大学 | Geologic body data visualization oriented feature-preservation quantifying method |
| CN103499819A (en) * | 2013-09-22 | 2014-01-08 | 中国科学院光电技术研究所 | Device and method for measuring angular offset and distance of target line of sight |
| CN103828232A (en) * | 2011-09-22 | 2014-05-28 | 伊尔索芙特有限公司 | Dynamic range control |
| US20160226554A1 (en) * | 2015-01-30 | 2016-08-04 | Trellisware Technologies, Inc. | Methods and systems for interference estimation via quantization in spread-spectrum systems |
| CN107688855A (en) * | 2016-08-12 | 2018-02-13 | 北京深鉴科技有限公司 | It is directed to the layered quantization method and apparatus of Complex Neural Network |
| CN108769677A (en) * | 2018-05-31 | 2018-11-06 | 宁波大学 | A kind of high dynamic range video dynamic range scalable encoding based on perception |
| WO2018209932A1 (en) * | 2017-05-17 | 2018-11-22 | 清华大学 | Multi-quantization depth binary feature learning method and device |
| US20190072662A1 (en) * | 2015-10-09 | 2019-03-07 | Zte Corporation | Method for transmitting a quantized value in a communication system |
| CN109495415A (en) * | 2018-10-12 | 2019-03-19 | 武汉邮电科学研究院有限公司 | Transmission method and link before digital mobile based on number cosine converting and segment quantization |
| RU2691588C1 (en) * | 2018-09-27 | 2019-06-14 | Федеральное государственное бюджетное образовательное учреждение высшего образования "Поволжский государственный технологический университет" | Analogue-to-digital and digital-to-analogue conversion method with non-uniform amplitude quantisation |
| US20190206034A1 (en) * | 2018-01-04 | 2019-07-04 | Boe Technology Group Co., Ltd. | Image enhancement method and device |
| CN110796622A (en) * | 2019-10-30 | 2020-02-14 | 天津大学 | Image bit enhancement method based on multi-layer characteristics of series neural network |
| CN110852964A (en) * | 2019-10-30 | 2020-02-28 | 天津大学 | Image bit enhancement method based on deep learning |
| CN110865728A (en) * | 2018-08-27 | 2020-03-06 | 苹果公司 | Force or touch sensing on mobile devices using capacitance or pressure sensing |
| CN110874625A (en) * | 2018-08-31 | 2020-03-10 | 杭州海康威视数字技术股份有限公司 | Deep neural network quantification method and device |
| US20200090601A1 (en) * | 2017-10-10 | 2020-03-19 | HKC Corporation Limited | Driving method and apparatus for display apparatus |
| CN111340692A (en) * | 2018-12-18 | 2020-06-26 | 北京长峰科威光电技术有限公司 | Infrared image dynamic range compression and contrast enhancement algorithm |
| US20210133278A1 (en) * | 2019-11-01 | 2021-05-06 | Samsung Electronics Co., Ltd. | Piecewise quantization for neural networks |
| CN113408704A (en) * | 2021-06-29 | 2021-09-17 | 深圳市商汤科技有限公司 | Data processing method, device, equipment and computer readable storage medium |
| CN113780301A (en) * | 2021-07-26 | 2021-12-10 | 天津大学 | Self-adaptive denoising machine learning application method for defending against attack |
-
2022
- 2022-04-15 CN CN202210398259.0A patent/CN114708180B/en active Active
Patent Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101854714A (en) * | 2010-05-13 | 2010-10-06 | 清华大学 | Method for achieving wireless communication timing coarse synchronization by using 1bit quantification and hard decision |
| CN103828232A (en) * | 2011-09-22 | 2014-05-28 | 伊尔索芙特有限公司 | Dynamic range control |
| CN103295208A (en) * | 2013-05-09 | 2013-09-11 | 浙江大学 | Geologic body data visualization oriented feature-preservation quantifying method |
| CN103499819A (en) * | 2013-09-22 | 2014-01-08 | 中国科学院光电技术研究所 | Device and method for measuring angular offset and distance of target line of sight |
| US20160226554A1 (en) * | 2015-01-30 | 2016-08-04 | Trellisware Technologies, Inc. | Methods and systems for interference estimation via quantization in spread-spectrum systems |
| US20190072662A1 (en) * | 2015-10-09 | 2019-03-07 | Zte Corporation | Method for transmitting a quantized value in a communication system |
| CN107688855A (en) * | 2016-08-12 | 2018-02-13 | 北京深鉴科技有限公司 | It is directed to the layered quantization method and apparatus of Complex Neural Network |
| WO2018209932A1 (en) * | 2017-05-17 | 2018-11-22 | 清华大学 | Multi-quantization depth binary feature learning method and device |
| US20200090601A1 (en) * | 2017-10-10 | 2020-03-19 | HKC Corporation Limited | Driving method and apparatus for display apparatus |
| US20190206034A1 (en) * | 2018-01-04 | 2019-07-04 | Boe Technology Group Co., Ltd. | Image enhancement method and device |
| CN108769677A (en) * | 2018-05-31 | 2018-11-06 | 宁波大学 | A kind of high dynamic range video dynamic range scalable encoding based on perception |
| CN110865728A (en) * | 2018-08-27 | 2020-03-06 | 苹果公司 | Force or touch sensing on mobile devices using capacitance or pressure sensing |
| CN110874625A (en) * | 2018-08-31 | 2020-03-10 | 杭州海康威视数字技术股份有限公司 | Deep neural network quantification method and device |
| RU2691588C1 (en) * | 2018-09-27 | 2019-06-14 | Федеральное государственное бюджетное образовательное учреждение высшего образования "Поволжский государственный технологический университет" | Analogue-to-digital and digital-to-analogue conversion method with non-uniform amplitude quantisation |
| CN109495415A (en) * | 2018-10-12 | 2019-03-19 | 武汉邮电科学研究院有限公司 | Transmission method and link before digital mobile based on number cosine converting and segment quantization |
| CN111340692A (en) * | 2018-12-18 | 2020-06-26 | 北京长峰科威光电技术有限公司 | Infrared image dynamic range compression and contrast enhancement algorithm |
| CN110852964A (en) * | 2019-10-30 | 2020-02-28 | 天津大学 | Image bit enhancement method based on deep learning |
| CN110796622A (en) * | 2019-10-30 | 2020-02-14 | 天津大学 | Image bit enhancement method based on multi-layer characteristics of series neural network |
| US20210133278A1 (en) * | 2019-11-01 | 2021-05-06 | Samsung Electronics Co., Ltd. | Piecewise quantization for neural networks |
| CN113408704A (en) * | 2021-06-29 | 2021-09-17 | 深圳市商汤科技有限公司 | Data processing method, device, equipment and computer readable storage medium |
| CN113780301A (en) * | 2021-07-26 | 2021-12-10 | 天津大学 | Self-adaptive denoising machine learning application method for defending against attack |
Non-Patent Citations (4)
| Title |
|---|
| WAN P等: ""High bit-depth image acquisition framework using embedded quantization bias"" * |
| WAN P等: ""High bit-precision image acquisition and reconstruction by planned sensor distortion"" * |
| 徐悦等: ""关于机器人导航目标点搜索路径模糊控制"" * |
| 欧阳慧明等: ""红外图像动态范围压缩算法研究综述"" * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114708180B (en) | 2023-05-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20200151914A1 (en) | Ai encoding apparatus and operation method of the same, and ai decoding apparatus and operation method of the same | |
| US7076114B2 (en) | Block boundary artifact reduction for block-based image compression | |
| JP6141295B2 (en) | Perceptually lossless and perceptually enhanced image compression system and method | |
| US7643688B2 (en) | Reducing artifacts in compressed images | |
| CN108476325B (en) | Media, method, and apparatus for high dynamic range color conversion correction | |
| CN1175846A (en) | Image enhancement method and circuit using quantized mean-matching histogram equalization | |
| CN110944176B (en) | Image frame noise reduction method and computer storage medium | |
| US4776029A (en) | Method of compressing image signals | |
| US6640017B1 (en) | Method and apparatus for adaptively sharpening an image | |
| US11494946B2 (en) | Data compression device and compression method configured to gradually adjust a quantization step size to obtain an optimal target quantization step size | |
| US10742986B2 (en) | High dynamic range color conversion correction | |
| CN111738951A (en) | Image processing method and device | |
| EP3557872A1 (en) | Method and device for encoding an image or video with optimized compression efficiency preserving image or video fidelity | |
| CN114708180B (en) | Bit depth quantization and enhancement method for predistortion image with dynamic range preservation | |
| CN112040231B (en) | Video coding method based on perceptual noise channel model | |
| KR100885441B1 (en) | Filtering method for block boundary region | |
| CN114697709B (en) | Video transmission method and device | |
| KR100598368B1 (en) | Filtering method for block boundary region | |
| CN116567286B (en) | Online live video processing method and system based on artificial intelligence | |
| KR100598369B1 (en) | Filtering method for block boundary region | |
| KR100524856B1 (en) | Filtering method block boundary region | |
| CN102905129B (en) | Distributed coding method of still image | |
| US10715772B2 (en) | High dynamic range color conversion correction | |
| CN115861089A (en) | Method and system for enhancing SF6 infrared image edge information | |
| CN116823970A (en) | Method for converting color image and gray image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |