TW201419861A - Tone mapping method of high dynamic range image - Google Patents

Abstract

There is provided a tone mapping method of high dynamic range image, which comprises the following steps: (A) acquiring a high dynamic range (HDR) image; (B) performing two processes to the HDR image, wherein the first one is to perform a fast screen type tone mapping process to the HDR image so that each processed pixel is outputted as PixelFG, and the second one is to perform a block-based operation of gradient compression-based algorithm so that each processed pixel is outputted as PixelBGC; and c performing a calculation to the pixel output PixelFG and the pixel output PixelBGC corresponding to each pixel and outputting the calculated result as the final result , wherein the calculation is αx PixelFG + β x PixelBGC where the values of α and β are in the range of 0 to 1 and α+βequals 1.

Description

Tone correspondence method for high dynamic range images

The invention relates to the technology of image processing, in particular to a method for matching the color of a high dynamic range image.

There are still many differences between the current camera and the real scene. One of the most important factors, even if these devices can support the dynamic range of brightness, is far less than the human eye can perceive. At present, the progress of photosensitive elements and the advancement of image processing technology have been observed. Both of these advances will effectively enhance the dynamic range of the camera and bring it closer to the real scene.

However, current displays are limited by their electronic components and still only support low dynamic range, so even in the field of imaging, high dynamic range can be obtained through advanced sensing components or through advanced image processing technology. Hereinafter referred to as HDR), however, these HDR images cannot be directly displayed correctly, and the Tone Mapping (TM) program must be used to compress the dynamic range before the original image can be correctly displayed on the display.

At present, the best-performing tone-corresponding algorithm is the Gradient Compression-Based (GC) algorithm, but since the GC algorithm involves calculating the gradient and the attenuation function, and the Poisson equation must be solved, the operation is very complicated. Just a static image takes a long time, so you can't respond to the instant video processing in pure software, but you must use a hardware accelerator to assist the operation. As a result, hard The increase in body cost and long processing time are not economical.

Because the GC algorithm is very complicated, it is not suitable to implement all the operations directly with the size of the screen. The current solution is to cut the image into blocks and then perform calculations for each block (hereafter referred to as Block-based GC) in order to maintain the hardware cost within an acceptable range. At present, regardless of whether there is overlap between blocks, the general solution is to seek the largest block under the hardware cost limit, so as to reduce the image quality caused by the block. However, no matter how the blocks are divided, the Block-based GC loses the original continuous color tone of the picture due to the independent operation of the block, which leads to the deterioration of image quality. This problem has not been solved yet. . Therefore, although the Block-based GC is inevitable, it also has the problem of poor image quality.

In addition, there is a simple calculation of the "Frame-Based" tone conversion technology, which does not require a large amount of hardware computing costs, such as Photographic (hereinafter referred to as PH) algorithm [1], or Fast Bilateral. The Filtering method [2], etc., does not have the problem of poor image quality, but still needs to use a certain amount of hardware computing resources, so it can be said that the Block-based GC has a long and complementary effect.

Since the full popularity of HDR images is almost an established trend, it is indispensable to display HDR image data instantly and quickly. Therefore, devices with instant and high-quality TM computing capabilities have considerable commercial potential, and the present invention is It is echoed by this trend.

[1] E. Reinhard, M. Stark, P. Shirley, and J. Ferwerda, “Photographic Tone Reproduction for Digital Images,” Proc Of SIGGRAPH, 2002.

[2] F. Durand and J. Dorsey, “Fast Bilateral Filtering for the Display of HDR Images,” ACM Transactions on Graphics, 21(3): 257-266, 2002.

The main object of the present invention is to provide a method for matching the tone of a high dynamic range image, which can be combined with a block-based GC and a frame-based tone conversion technique. With simplified calculations and improved image quality, it has the effect of reducing hardware requirements and processing quickly.

In order to achieve the foregoing object, a method for color tone corresponding to a high dynamic range image according to the present invention includes the following steps: A) obtaining a high dynamic range (HDR) image, the HDR image is composed of a plurality of screens, each of which The picture is composed of a plurality of pixels; B) performing two kinds of processing on the HDR image, one of which is a Fast Global Tone Mapping process for the HDR image, and each pixel processed by the HDR image The output is Pixel _FG , in which the pixel output Pixel _FG corresponding to each pixel is not necessarily the same; the second is to cut the HDR image into multiple blocks and then perform a block-based gradient compression algorithm for each block. (Block-based GC) operation, and each pixel output processed is Pixel _BGC , in which the pixel output Pixel _BGC corresponding to each pixel is not necessarily the same; and C) the pixel corresponding to each pixel is output Pixel _FG the pixel output with a pixel _BGC operation, this operation content is _{α × pixel FG + β × pixel} BGC, wherein the value beta] [alpha] and is between 0 and 1, and α + β = 1; i.e., the result of the calculation The final output.

Preferably, in step C), the values of α and β are dynamically adjusted depending on the difference of the block or the difference of the picture, but it is still necessary to maintain the values of α and β between 0 and 1, and α+β= 1.

Preferably, in step B), the fast picture type tone correspondence processing refers to a frame-based processing method, and is a method for directly processing the entire picture, and the operation complexity thereof is not More than the block-based GC-based gradient compression algorithm (Block-based GC).

Preferably, the fast-screen tone matching processing is one of two technologies selected from the group consisting of a Photographic Tone Mapping (PH) or a Fast Bilateral Filtering.

In order to explain the technical features of the present invention in detail, the following preferred embodiments are described below with reference to the drawings.

As shown in the first figure, a method for color tone correspondence of a high dynamic range image according to a preferred embodiment of the present invention has the following steps:

A) A high dynamic range (HDR) image is obtained, which is composed of a plurality of pictures, each of which is composed of a plurality of pixels.

B) performing two processes on the HDR image, one of which is to perform a Fast Global Tone Mapping process on the HDR image, and each pixel processed by the pixel is Pixel _FG , wherein each pixel is The corresponding pixel output Pixel _{FG is} not necessarily the same; the second is to block the HDR image into multiple blocks and then perform a block-based GC-based operation for each block, and The output of each pixel processed is Pixel _BGC , in which the pixel output Pixel _BGC corresponding to each pixel is not necessarily the same.

In the present embodiment, the fast picture type tone correspondence processing refers to a frame-based processing method, and is a method for directly processing the entire picture, and the operation complexity thereof is not more than the block. A basic block-based GC. For example, the fast-screen tone matching processing is one of a technique of selecting either a Photographic Tone Mapping (PH) or a Fast Bilateral Filtering.

C) performing an operation on the pixel output Pixel _FG corresponding to each pixel and the pixel output Pixel _BGC , the operation content is α×Pixel _FG +β×Pixel _BGC , wherein the values of α and β are between 0 and 1 Between, and α + β = 1; the result of the operation is the final output. In this embodiment, the values of α and β are dynamically adjusted depending on the block or the difference in the picture, but it is still necessary to maintain the values of α and β between 0 and 1, and α+β=1.

Through the above steps, the advantage of the fast picture tone processing method can be obtained: the whole picture is processed, but not divided into blocks, thereby maintaining the picture quality; on the other hand, the block-based gradient compression algorithm can also be obtained. (Block-based GC) Advantages: Fast processing with hardware limitations, reducing hardware requirements and reducing costs. Together with the weight distribution (ie the ratio of α and β), the most suitable image output can be adjusted. By this, this The technique provided by the invention solves the limitation on the hardware cost in the prior art, and also solves the problem of using a block-based GC to cause a deterioration in picture quality.

It can be seen that the present invention achieves the effects of simplifying the calculation and improving the image quality, thereby reducing the hardware requirement and the fast processing effect.

The first figure is a schematic view of the operation of a preferred embodiment of the present invention.

Claims (4)

A method for matching a tone of a high dynamic range image includes the following steps: A) obtaining a high dynamic range (HDR) image, the HDR image being composed of a plurality of pictures, each of the pictures being composed of a plurality of pixels; The HDR image is subjected to two processes, one of which is a Fast Global Tone Mapping process, and each pixel processed is Pixel _FG , wherein each pixel corresponds to The pixel output Pixel _{FG is} not necessarily the same; the second is to block the HDR image into a plurality of blocks and then perform a block-based GC-based block operation for each block, and Each pixel output processed is Pixel _BGC , wherein the pixel output Pixel _BGC corresponding to each pixel is not necessarily the same; and C) an operation is performed on the pixel output Pixel _FG corresponding to each pixel and the pixel output Pixel _BGC , The operation content is α×Pixel _FG +β×Pixel _BGC , where the values of α and β are between 0 and 1, and α+β=1; the result of the operation is the final output result. According to the color tone corresponding method of the high dynamic range image according to claim 1, wherein in step C), the values of α and β are dynamically adjusted according to different blocks or different screens, but it is still necessary to maintain α. And the value of β is between 0 and 1, and α + β = 1. According to the color tone corresponding method of the high dynamic range image according to the first aspect of the patent application, in the step B), the fast picture type tone corresponding processing refers to a frame-based processing method. and It is a way to directly process the entire picture, and its computational complexity does not exceed the block-based GC-based gradient compression algorithm (Block-based GC). According to the color tone corresponding method of the high dynamic range image described in claim 3, wherein the fast picture type tone corresponding processing is selected from a Photographic Tone Mapping (PH) or a fast bilateral filtering method ( Fast Bilateral Filtering) One of the two technologies.