JND coefficient value range method and system

Abstract

The EOTF brightness adjustment method, JND coefficient value range method and system relate to the field of image display technology, and solve the problems of perceived contrast loss and artifacts in low-brightness scenes caused by the existing EOTF under 10-bit PQ quantization. Specifically: based on the value range of the JND coefficient, combined with the adaptive adjustment factor and the Barten function, the value of the JND coefficient is determined; based on the value of the JND coefficient, the brightness and JND coefficient relationship function is repeatedly input until the peak brightness is obtained, and then the operation is terminated. The EOTF brightness adjustment method described in the present invention is applicable to any changes in peak brightness and parameter values.

Description

Technical Field

[0001] The present invention relates to the field of image display technology, and in particular to a method and system for determining a value range of a JND coefficient. Background Art

[0002] HDR (High Dynamic Range) image display technology has been widely used in recent years. It provides users with a more realistic visual experience by providing higher brightness, richer colors and higher contrast. For image acquisition and display, the use of transfer functions has a long history, which reflects the mapping relationship between actual brightness and digital signals. The acquisition device converts the actual brightness into an encoded digital signal through OETF (Otoelectronic Transfer Function), and the display terminal converts the digital signal into an optical signal through EOTF (Electro-Optical Transfer Function) to present the picture. During the display process, the design of EOTF will directly affect the size of quantization noise. This noise appears as a color band and artifact similar to "contour lines", causing obvious brightness or color faults in the originally smooth transition position of the picture. Figure 1 and Figure 2 As can be seen from the comparison, this illustrates an extreme case of this type of noise, where pseudo contours appear in smooth areas due to the mismatch between the brightness mapping relationship of the EOTF and the perception of the human eye.

[0003] In the SDR (standard dynamic range) system, a transfer function commonly known as "gamma" is used. The nonlinear mapping achieved by the transfer function is called "gamma correction". The "gamma" values ​​used for display screens of different materials are also different, usually between 2.2 and 3.0. The two standards ITU-R BT.709 and ITU-R BT.1886 define OETF with gamma = 0.45 and EOTF with gamma = 2.4 respectively. However, "gamma" shows great deficiencies when used for HDR display.

[0004] In order to adapt to the HDR system and explore the ideal relationship between the brightness response of the display and the digital signal, a lot of related work has been done before. The British BBC Broadcasting Corporation and the Japanese Broadcasting Association NHK jointly developed HLG (Hybrid Log-Gamma) to meet the needs of HDR technology in digital TV broadcasting. The characteristic of HLG is that it is backward compatible with SDR (standard dynamic range) content. This means that HLG content can present a richer and more moving picture effect even when played on existing SDR display devices. In addition, Scott Miller and others proposed an EOTF called PQ (perceptual quantizer) based on the Barten function (contrast sensitivity function, also known as CSF). PQ can quantize 10 through a 12-bit integer bit depth. -6 ~10000cd / m 2brightness range. Considering the compatibility issues with current consumer electronic products, 10-bit PQ is currently used more as the EOTF of HDR display screens to save resources. In 2014, PQ was adopted as a standard by SMPTE (Society of Motion Picture and Television Engineers) and recorded in SMPTE ST-2084. Currently, PQ and HLG have been standardized by ITU (International Telecommunication Union) and recorded in ITU-RBT.2100. However, although 10-bit PQ can already meet the needs of most scenes, there is still a certain risk of distortion when using 10-bit PQ when the dynamic range is high enough, especially when displaying low-brightness scenes, the visible contrast information cannot be fully retained.

[0005] In summary, during the HDR image display process, the existing EOTF will cause perceptual contrast loss and artifacts in low-brightness scenes under 10-bit PQ quantization. Summary of the invention

[0006] The present invention solves the problems of perceived contrast loss and artifacts in low-brightness scenes caused by the existing EOTF under 10-bit PQ quantization.

[0007] The EOTF brightness adjustment method of the present invention is specifically as follows:

[0008] Based on the value range of the JND coefficient, combined with the adaptive adjustment factor and the Barten function, the value of the JND coefficient is determined;

[0009] The value based on the JND coefficient is repeatedly input into the brightness and JND coefficient relationship function until the peak brightness is obtained, and then the operation is terminated.

[0010] Further, in one embodiment of the present invention, the value of the JND coefficient is:

[0011] f i =S(L i )·α·L i -β 0 <i<1023;

[0012] Among them, f is the JND coefficient, S is the Barten function, L is the brightness, α and β are both adjustment parameters, and i is the i-th brightness.

[0013] The JND coefficient value range method described in the present invention is used to implement the EOTF brightness adjustment method described in the above method, specifically:

[0014] Based on the relationship function between RC and ST and the Barten function, the value of the right boundary of the JND coefficient is derived;

[0015] Determine the value of the left boundary of the JND coefficient based on the MDC threshold;

[0016] The value range of the JND coefficient is between the value of the right boundary of the JND coefficient and the value of the left boundary of the JND coefficient.

[0017] Furthermore, in one embodiment of the present invention, the relationship function between RC and ST is RC <ST。

[0018] Furthermore, in one embodiment of the present invention, the calculation formula of RC is:

[0019]

[0020] Among them, f is the JND coefficient, S is the Barten function, and L is the brightness.

[0021] Further, in one embodiment of the present invention, the value of the right boundary of the JND coefficient is:

[0022]

[0023] Among them, S is the Barten function, L is the brightness, and ST is the Schreiber threshold.

[0024] Furthermore, in one embodiment of the present invention, the calculation formula of the MDC threshold is:

[0025]

[0026] Where, MDC is the MDC threshold, m t is the modulation threshold.

[0027] Further, in one embodiment of the present invention, the value of the left boundary of the JND coefficient is

[0028] The EOTF brightness adjustment system of the present invention is specifically:

[0029] Based on the value range of the JND coefficient, combined with the adaptive adjustment factor and the Barten function, the value of the JND coefficient is determined;

[0030] The value based on the JND coefficient is repeatedly input into the brightness and JND coefficient relationship function until the peak brightness is obtained, and then the operation is terminated.

[0031] The JND coefficient value range system of the present invention is used to implement the EOTF brightness adjustment system, specifically:

[0032] Based on the relationship function between RC and ST and the Barten function, the value of the right boundary of the JND coefficient is derived;

[0033] The value of the left boundary of the JND coefficient is determined based on the threshold of the Barten function;

[0034] The value range of the JND coefficient is between the value of the right boundary of the JND coefficient and the value of the left boundary of the JND coefficient.

[0035] The present invention solves the problem of perceived contrast loss and artifacts in low-brightness scenes caused by the existing EOTF under 10-bit PQ quantization. Specific beneficial effects include:

[0036] 1. The EOTF brightness adjustment method described in the present invention solves the problems of perceived contrast loss and artifacts in low-brightness scenes caused by the existing EOTF under 10-bit PQ quantization during HDR image display. In order to solve the above technical problems, the present invention optimizes the brightness mapping relationship to make the quantization intervals in different brightness ranges more uniform, thereby improving the perceived contrast of the overall image. The quantization intervals in the low-brightness range are finely adjusted through adaptive adjustment factors to eliminate artifacts in low-brightness scenes. Based on the visual characteristics of the human eye, a reasonable brightness mapping relationship is designed to make the brightness reproduction in HDR more realistic and natural. That is, the present invention solves the problems of perceived contrast loss and artifacts in low-brightness scenes caused by the existing EOTF by improving perceived contrast, eliminating low-brightness artifacts, and optimizing brightness mapping relationships.

[0037] 2. The JND coefficient value range method of the present invention determines the value of the right boundary of the JND coefficient and the value of the left boundary of the JND coefficient respectively based on the relationship function between RC and ST and the Barten function, thereby determining the value range of the JND coefficient. The value range is combined with the adaptive adjustment factor and the Barten function to determine the value of the JND coefficient, which is a key factor for adjusting the brightness of the EOTF;

[0038] The EOTF brightness adjustment method described in the present invention is applicable to any changes in peak brightness and parameter values. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0040] Figure 1 It is the normal picture picture described in the background technology;

[0041] Figure 2 It is the noise picture diagram described in the background technology;

[0042] Figure 3 is the MDC threshold curve diagram described in Implementation Mode 1;

[0043] Figure 4 It is a position diagram of the PQ curve described in the first implementation mode on the Barten function. DETAILED DESCRIPTION

[0044] The following will clearly and completely describe various embodiments of the present invention in conjunction with the accompanying drawings. The embodiments described with reference to the accompanying drawings are exemplary and intended to be used to explain the present invention, but should not be understood as limiting the present invention.

[0045] Implementation method 1: The EOTF brightness adjustment method described in this implementation method is specifically as follows:

[0046] Based on the value range of the JND coefficient, combined with the adaptive adjustment factor and the Barten function, the value of the JND coefficient is determined;

[0047] The value based on the JND coefficient is repeatedly input into the brightness and JND coefficient relationship function until the peak brightness is obtained, and then the operation is terminated.

[0048] In this implementation manner, the value of the JND coefficient is:

[0049] f i =S(L i )·α·L i -β 0 <i<1023;

[0050] Among them, f is the JND coefficient, S is the Barten function, L is the brightness, α and β are both adjustment parameters, and i is the i-th brightness.

[0051] In order to better describe the EOTF brightness adjustment method described in this embodiment, several terms and the derivation process of PQ are introduced here:

[0052] 1. Barten function: a psychophysical model used to describe the sensitivity of the human eye to brightness and spatial frequency. The Barten function expression is as follows:

[0053]

[0054] Among them, S is the Barten function, u is the spatial frequency, L is the brightness, m t is the modulation threshold.

[0055] 2. m t is the inverse of the Barten function, which can also be defined by the following formula:

[0056]

[0057] Among them, i is the i-th brightness.

[0058] Based on the above formula, the brightness relationship between any two levels can be obtained as follows:

[0059]

[0060] M t The weighting with the JND (Just Noticeable Difference) coefficient f can adjust the step accuracy of each brightness change. When f = 1, it is 1 JND, and its expression is as follows:

[0061]

[0062] 3. When the spatial frequency u is the spatial frequency that the human eye perceives most clearly at the current brightness under a fixed brightness, it means that each level of brightness L corresponds to a fixed spatial frequency u, and the Barten function becomes a single-valued function of brightness L. After the starting brightness L is determined, the starting brightness L is substituted into (1) to obtain the corresponding modulation threshold m t , and then m t Substituting into (4), we can obtain the next level of brightness L with an interval of JND coefficient f. By repeatedly iterating and solving until the brightness L is greater than or equal to 10000nits, the number of brightness L levels obtained is the required bit depth. The results show that the EOTF designed based on the Barten function needs to be constructed with 12 bits. When the brightness L is 10 -6 ~10000cd / m 2 When the JND coefficient f = 0.9, the accuracy of each brightness L is 0.9 JNDs. Fitting the 12-bit brightness L into a smooth curve is the PQ curve, such as Figure 4 The figure shows the position of the PQ curve on the Barten function, where the figure shows the relationship between the Barten function and the spatial frequency u and the brightness L, that is, S(u,L).

[0063] However, due to many factors, the 12-bit depth is currently compressed to 10-bit. However, the 10-bit PQ quantization will cause the loss of perceptual contrast and artifacts in low-brightness scenes. The technical problem is described in detail as follows:

[0064] like Figure 3 As shown in the figure, two MDC (minimum detectable contrast) threshold curves in ITU-R BT.2246 are shown, where the horizontal axis represents L (brightness) in cd / m 2The vertical axis represents the minimum reproducible contrast. The two MDC threshold curves are the dark dashed line representing the Barten function and the light dashed line representing ST (Schreiber Threshold, a name in the field of psychotherapy). The other three curves represent 10-bit gamma3.0 10000cd / m from top to bottom. 2 , gamma3.0 10000cd / m 2 , 10-bit PQ and 10-bit HLG 4000cd / m 2 . Generally, when RC (reproducible contrast) is below the Barten function, it is believed that no one can perceive brightness changes, and when RC is above ST, there is a greater possibility of artifacts. It can be observed from the figure that although the three curves show high accuracy in the highlight range, they have exceeded the ST threshold when displaying dark scenes, which means that there is a risk of contour artifacts in the corresponding brightness range.

[0065] Therefore, in HDR image display, existing EOTFs, such as PQ and HLG, have several technical problems when processing 10-bit quantization, including the following aspects:

[0066] 1) Perceptual contrast loss: Under 10-bit quantization conditions, the existing EOTF has redundancy in the quantization interval distribution in different brightness ranges, resulting in a loss of perceptual contrast;

[0067] 2) Artifacts in low-brightness scenes: Since the quantization interval of the existing EOTF in the low-brightness range is too large, artifacts are easily generated when displaying low-brightness images.

[0068] These artifacts not only affect the image quality, but may also have a negative impact on the user's viewing experience. In order to solve the above technical problems, this embodiment proposes an EOTF brightness adjustment method, which is specifically:

[0069] In order to make PQ adapt to the peak brightness of 10000nits, a JND coefficient f fixed to 0.9 is used in its design. If you want to compress the 12-bit depth to 10-bit, a relatively ideal solution is to adaptively adjust the JND coefficient f according to different brightness. Therefore, this embodiment proposes an EOTF brightness adjustment method, which is as follows:

[0070] First, we need to discuss the value range of the JND coefficient f. Simply put, the value range of the JND coefficient f is between the value of the right boundary of the JND coefficient f and the value of the left boundary of the JND coefficient f.

[0071] The experimental results of Dolby Laboratories' subjective test show that 0.1cd / m2 and 600cd / m 2 Meet the black level and diffuse white maximum level requirements of 50% of viewers. 0.005cd / m 2 and 3000cd / m 2 The black level and the maximum level of diffuse white can meet the requirements of 90% of viewers. Based on the above conclusions, taking into account the 100cd / m 2 For the reference white brightness of ITU-R BT.1886 standard, 10 -6 ~10000cd / m 2 It is divided into 7 brightness ranges. 2 Set as the dividing line between dark and night scenes, 0.1cd / m 2 Set as the dividing line between night and dim scenes, 3000cd / m 2 Set as the dividing line between bright and high-brightness scenes, 600cd / m 2 Set as the boundary between bright and medium-bright scenes. Combined with the value range of the JND coefficient f, define the adaptive adjustment factor. The final value of the JND coefficient f is determined by the adaptive adjustment factor and the Barten function, and the expression is given by (5):

[0072] f i =S(L i )·α·L i -β 0 <i<1023; (5)

[0073] Among them, f is the JND coefficient, S is the Barten function, L is the brightness, α and β are adjustment parameters, and i is the i-th brightness, that is, f i is the value of the JND coefficient f at the ith brightness level, S(L i ) is f i is the contrast sensitivity of the human eye at the i-th brightness level.

[0074] The values ​​of adjustment parameters α and β in different brightness L intervals are given by (6):

[0075]

[0076] Once the JND coefficient f is determined, f i Repeat (4) to get the next level of brightness L i+1 , until it reaches 10000nits, the brightness L reaches its peak when the 12-bit depth is compressed to 10-bit.

[0077] Therefore, this implementation designs a new brightness mapping relationship based on the Barten function, so as to better preserve contrast information and improve the display effect of HDR content.

[0078] Implementation method 2: The JND coefficient value range method described in this implementation method is used to implement the EOTF brightness adjustment method described in implementation method 1, specifically:

[0079] Based on the relationship function between RC and ST and the Barten function, the value of the right boundary of the JND coefficient is derived;

[0080] Determine the value of the left boundary of the JND coefficient based on the MDC threshold;

[0081] The value range of the JND coefficient is between the value of the right boundary of the JND coefficient and the value of the left boundary of the JND coefficient.

[0082] In this embodiment, the relationship function between RC and ST is RC <ST。

[0083] In this implementation manner, the calculation formula of RC is:

[0084]

[0085] Among them, f is the JND coefficient, S is the Barten function, and L is the brightness.

[0086] In this implementation manner, the value of the right boundary of the JND coefficient is:

[0087]

[0088] Among them, S is the Barten function, L is the brightness, and ST is the Schreiber threshold.

[0089] In this implementation, the calculation formula of the threshold of the Barten function is:

[0090]

[0091] Where, MDC is the MDC threshold, m t is the modulation threshold.

[0092] In this implementation, the left boundary value of the JND coefficient is

[0093] In order to implement the EOTF brightness adjustment method described in Implementation 1 and determine the value of the JND coefficient f, this implementation proposes a JND coefficient value range method, specifically:

[0094] To ensure that RC is always below ST, it is necessary to satisfy RC < ST. Substituting formula (4) into the calculation formula of RC gives formula (7):

[0095]

[0096] After derivation, is the value of the right boundary of the JND coefficient f.

[0097] If the result is to be meaningful, the JND coefficient must be made Then it is necessary to find the relationship between the JND coefficient f, the spatial frequency u, and the luminance L. Therefore, first take the first-order partial derivative of the Barten function with respect to the spatial frequency u, and the result is as in (8):

[0098]

[0099] Let Then the function of the spatial frequency u with respect to the luminance L when the Barten function reaches the extreme value can be solved, and substituting it into formula (7) to obtain the value of the right boundary of the JND coefficient f.

[0100] In addition, to save the bit depth of the code value, it is necessary to determine the value of the left boundary of a JND coefficient f. When the Barten function was established, a sine-wave grating was used as the experimental object because it can provide continuous contrast changes and uniform spatial frequency characteristics. However, for the MDC threshold, it is more reasonable to use a sawtooth wave, which can provide sharper edges and a gradually changing luminance pattern. Since the relative amplitude of the fundamental frequency component of the Fourier series expansion of the sawtooth wave is equal to Therefore, the calculation method of the MDC threshold (also known as the Barten threshold) is as in (9):

[0101]

[0102] Considering that no one can perceive the luminance change below the MDC threshold, and the 10-bit depth cannot meet its requirements. Therefore, the detectable contrast is placed between ST and the MDC threshold, and π / 2 is the value of the left boundary of the JND coefficient f.

[0103] Based on this, the final value range of the JND coefficient f should satisfy

[0104] Based on the value range of the JND coefficient f, combined with the adaptive adjustment factor and the Barten function, the value of the JND coefficient f can be determined, providing a necessary basis for solving the problems of the existing technology.

[0105] Embodiment 3. The EOTF luminance adjustment system described in this embodiment is specifically as follows:

[0106] Based on the value range of the JND coefficient, combined with the adaptive adjustment factor and the Barten function, the value of the JND coefficient is determined;

[0107] The value based on the JND coefficient is repeatedly input into the brightness and JND coefficient relationship function until the peak brightness is obtained, and then the operation is terminated.

[0108] Implementation method 4: The JND coefficient value range system described in this implementation method is used to implement the EOTF brightness adjustment system described in implementation method 3, specifically:

[0109] Based on the relationship function between RC and ST and the Barten function, the value of the right boundary of the JND coefficient is derived;

[0110] The value of the left boundary of the JND coefficient is determined based on the threshold of the Barten function;

[0111] The value range of the JND coefficient is between the value of the right boundary of the JND coefficient and the value of the left boundary of the JND coefficient.

[0112] The JND coefficient value range method and system proposed in the present invention are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. JND coefficient value range method, the method is used to implement the EOTF brightness adjustment method, characterized in that: Specifically: Based on the relationship function between reproducible contrast and Schreiber Threshold and Barten function, the value of the right boundary of JND coefficient is derived; Determine the value of the left boundary of the JND coefficient based on the MDC threshold; The value range of the JND coefficient is between the value of the right boundary of the JND coefficient and the value of the left boundary of the JND coefficient; EOTF brightness adjustment method, specifically: Based on the value range of the JND coefficient, combined with the adaptive adjustment factor and the Barten function, the value of the JND coefficient is determined; The value based on the JND coefficient is repeatedly input into the brightness and JND coefficient relationship function until the peak brightness is obtained, and then the operation is terminated.

2. The JND coefficient value range method according to claim 1, characterized in that: The value of the JND coefficient is: f i =S(L i )·a·L i -β 0 <i<1023; Among them, f is the JND coefficient, S is the Barten function, L is the brightness, α and β are both adjustment parameters, and i is the i-th brightness.

3. The JND coefficient value range method according to claim 1, characterized in that: The relationship function between the reproducible contrast and the Schreiber Threshold is: <Schreiber Threshold。 4. The JND coefficient value range method according to claim 1, characterized in that: The calculation formula of the reproducible contrast is: Among them, f is the JND coefficient, S is the Barten function, and L is the brightness.

5. The JND coefficient value range method according to claim 1, characterized in that: The value of the right boundary of the JND coefficient is: Among them, S is the Barten function, L is the brightness, and ST is the Schreiber threshold.

6. The JND coefficient value range method according to claim 1, characterized in that: The calculation formula of the MDC threshold is: Where, MDC is the MDC threshold, m t is the modulation threshold.

7. The JND coefficient value range method according to claim 1, characterized in that: The left boundary value of the JND coefficient is 8. A JND coefficient value range system, the system being used to implement an EOTF brightness adjustment method, characterized in that: Specifically: Based on the relationship function between reproducible contrast and Schreiber Threshold and Barten function, the value of the right boundary of JND coefficient is derived; Determine the value of the left boundary of the JND coefficient based on the MDC threshold; The value range of the JND coefficient is between the value of the right boundary of the JND coefficient and the value of the left boundary of the JND coefficient; EOTF brightness adjustment method, specifically: Based on the value range of the JND coefficient, combined with the adaptive adjustment factor and the Barten function, the value of the JND coefficient is determined; The value based on the JND coefficient is repeatedly input into the brightness and JND coefficient relationship function until the peak brightness is obtained, and then the operation is terminated.

Images