WO2023045907A1 - 多曝光图像处理方法、装置及降噪电路 - Google Patents

多曝光图像处理方法、装置及降噪电路 Download PDF

Info

Publication number
WO2023045907A1
WO2023045907A1 PCT/CN2022/119813 CN2022119813W WO2023045907A1 WO 2023045907 A1 WO2023045907 A1 WO 2023045907A1 CN 2022119813 W CN2022119813 W CN 2022119813W WO 2023045907 A1 WO2023045907 A1 WO 2023045907A1
Authority
WO
WIPO (PCT)
Prior art keywords
pixel
noise reduction
point
exposure
window
Prior art date
Application number
PCT/CN2022/119813
Other languages
English (en)
French (fr)
Inventor
李彦良
Original Assignee
Oppo广东移动通信有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Oppo广东移动通信有限公司 filed Critical Oppo广东移动通信有限公司
Publication of WO2023045907A1 publication Critical patent/WO2023045907A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • 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
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/70—Circuitry for compensating brightness variation in the scene
    • H04N23/76—Circuitry for compensating brightness variation in the scene by influencing the image signals
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N5/00—Details of television systems
    • H04N5/222—Studio circuitry; Studio devices; Studio equipment
    • H04N5/262—Studio circuits, e.g. for mixing, switching-over, change of character of image, other special effects ; Cameras specially adapted for the electronic generation of special effects
    • 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
    • G06T2207/20172—Image enhancement details
    • G06T2207/20208—High dynamic range [HDR] image processing
    • 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
    • G06T2207/20212—Image combination
    • G06T2207/20221—Image fusion; Image merging

Definitions

  • Embodiments of the present application relate to image processing technologies, and in particular to a multi-exposure image processing method, device and noise reduction circuit.
  • High Dynamic Range, HDR High Dynamic Range
  • DOL Digital Overlap
  • the specific method is to make long exposure, medium exposure and short exposure respectively for the same scene , three images are collected, and then these three images are fused, and finally a high dynamic range image is obtained.
  • DOL Digital Overlap
  • the specific method is to make long exposure, medium exposure and short exposure respectively for the same scene , three images are collected, and then these three images are fused, and finally a high dynamic range image is obtained.
  • DOL Digital Overlap
  • Embodiments of the present application are expected to provide a multi-exposure image processing method, device and noise reduction circuit.
  • a multi-exposure image processing method including:
  • a multi-exposure image processing device including:
  • An acquisition module configured to acquire at least two exposure images of the target scene captured at different exposure times
  • a noise reduction module configured to determine, based on the exposure time of the at least two exposure images, the size of a pixel window for performing noise reduction processing on the at least two exposure images; performing noise reduction processing on the at least two exposed images respectively;
  • the fusion module is configured to perform fusion processing on the at least two exposure images after noise reduction processing, so as to obtain a high dynamic range image of the target scene.
  • a noise reduction circuit in a third aspect, includes:
  • a pixel window construction circuit configured to load the pixel window of the exposure image according to the size of the pixel window of the exposure image performing noise reduction processing
  • An odd-numbered row extraction circuit configured to extract odd-numbered row pixels from the pixel window loaded by the pixel window construction circuit
  • An even-numbered row extraction circuit configured to extract even-numbered row pixels from the pixel window loaded by the pixel window construction circuit
  • a pixel matrix construction circuit configured to construct a pixel matrix according to the pixels in the odd rows or the pixels in the even rows;
  • the noise reduction operation circuit is configured to perform a noise reduction operation on the pixel matrix to output noise-reduced pixels.
  • a multi-exposure image processing chip is provided, and the chip includes the noise reduction circuit described in the aforementioned third aspect.
  • an electronic device in a fifth aspect, includes: an image acquisition device and the aforementioned multi-exposure image processing chip.
  • a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the aforementioned method are implemented.
  • a computer program product including a plurality of instructions, and when the instructions are executed by a computing device, the steps of the foregoing method are implemented.
  • FIG. 1 is a schematic flow diagram of a multi-exposure image processing method in an embodiment of the present application
  • FIG. 2 is a schematic flow diagram of a noise reduction processing method in an embodiment of the present application
  • FIG. 3 is a schematic diagram of the first size of the pixel window in the embodiment of the present application.
  • Fig. 4 is the first schematic diagram of the Gr point pixel matrix in the embodiment of the present application.
  • Fig. 5 is the first schematic diagram of the Gb point pixel matrix in the embodiment of the present application.
  • Fig. 6 is the first schematic diagram of the pixel matrix of point R in the embodiment of the present application.
  • Fig. 7 is the first schematic diagram of the pixel matrix at point B in the embodiment of the present application.
  • FIG. 8 is a schematic diagram of the second size of the pixel window in the embodiment of the present application.
  • FIG. 9 is a second schematic diagram of the Gr point pixel matrix in the embodiment of the present application.
  • Fig. 10 is the second schematic diagram of the Gb point pixel matrix in the embodiment of the present application.
  • FIG. 11 is a second schematic diagram of the pixel matrix at point R in the embodiment of the present application.
  • FIG. 12 is a second schematic diagram of the pixel matrix at point B in the embodiment of the present application.
  • FIG. 13 is a schematic diagram of the composition and structure of the multi-exposure image processing device in the embodiment of the present application.
  • FIG. 14 is a schematic diagram of the composition and structure of the noise reduction circuit in the embodiment of the present application.
  • FIG. 15 is a schematic diagram of the first component structure of the noise reduction operation circuit in the embodiment of the present application.
  • FIG. 16 is a schematic diagram of the second composition structure of the noise reduction operation circuit in the embodiment of the present application.
  • FIG. 17 is a schematic diagram of the composition and structure of the multi-exposure image processing chip in the embodiment of the present application.
  • FIG. 18 is a schematic diagram of the composition and structure of the electronic device in the embodiment of the present application.
  • the embodiment of the present application provides a multi-exposure image processing method, including:
  • the at least two exposure images are Bayer format images
  • performing noise reduction processing on the at least two exposure images based on the pixel window includes:
  • the sum and difference after noise reduction are averaged to obtain the second pixel of the center window after noise reduction.
  • the first pixel is the green point Gr point or the green point Gb point among the four Bayer format pixel points
  • the second pixel point is the red point R point or the blue point B point among the four Bayer format pixel points.
  • the noise reduction operation includes:
  • the method also includes:
  • the method also includes:
  • the noise reduction process is performed on the new pixel window until the noise reduction of all pixels of the exposure image is completed.
  • the size of the pixel window includes width and height
  • the size of the pixel window for performing noise reduction processing on the at least two exposure images is respectively determined based on the exposure time of the at least two exposure images include:
  • the at least two exposure images include a first exposure image captured at a first exposure time and a second exposure image captured at a second exposure time, and the method further includes:
  • the embodiment of the present application also provides a multi-exposure image processing device, including:
  • An acquisition module configured to acquire at least two exposure images of the target scene captured at different exposure times
  • a noise reduction module configured to determine, based on the exposure time of the at least two exposure images, the size of a pixel window for performing noise reduction processing on the at least two exposure images; performing noise reduction processing on the at least two exposed images respectively;
  • the fusion module is configured to perform fusion processing on the at least two exposure images after noise reduction processing, so as to obtain a high dynamic range image of the target scene.
  • the embodiment of the present application also provides a noise reduction circuit, and the noise reduction circuit includes:
  • a pixel window construction circuit configured to load the pixel window of the exposure image according to the size of the pixel window for performing noise reduction processing on the exposure image
  • An odd-numbered row extraction circuit configured to extract odd-numbered row pixels from the pixel window loaded by the pixel window construction circuit
  • An even-numbered row extraction circuit configured to extract even-numbered row pixels from the pixel window loaded by the pixel window construction circuit
  • a pixel matrix construction circuit configured to construct a pixel matrix according to the pixels in the odd rows or the pixels in the even rows;
  • the noise reduction operation circuit is configured to perform a noise reduction operation on the pixel matrix to output noise-reduced pixels.
  • the noise reduction operation circuit includes:
  • the absolute value subcircuit is configured to obtain an absolute value for each point in the pixel matrix
  • the square root subcircuit is configured to take the square root of the absolute value of each point in the pixel matrix to obtain the square root result sqrt_abs_cen of the center point of the pixel matrix and the square root result sqrt_abs_ref of the non-central point of the pixel matrix;
  • the comparison subcircuit is configured to compare the ABSDIFF with a preset threshold, and determine the target non-central points and the number of target non-central points whose ABSDIFF is smaller than the first threshold;
  • the first averaging sub-circuit is configured to accumulate all target non-central points and divide by the number of target non-central points to obtain denoised pixels.
  • the noise reduction operation circuit also includes:
  • the threshold configuration subcircuit is configured to obtain the weight and offset of the register configuration; multiply the square root result sqrt_abs_cen of the center point by the weight to obtain a product; add the product to the offset, get the first threshold.
  • the pixel matrix construction circuit includes: a Gr point pixel matrix construction circuit and a Gb point pixel matrix construction circuit;
  • the Gr point pixel matrix construction circuit is configured to construct a Gr point pixel matrix according to the Gr point pixels in the odd-numbered rows of pixels;
  • the Gb-dot pixel matrix construction circuit is configured to construct a Gb-dot pixel matrix according to the Gb-dot pixels in the even-numbered rows of pixels.
  • the pixel matrix construction circuit includes: an R point pixel matrix construction circuit and a B point pixel matrix construction circuit;
  • the R point pixel matrix construction circuit is configured to construct an R point pixel matrix according to the R point pixels in the odd row of pixels;
  • the B-point pixel matrix construction circuit is configured to construct the B-point pixel matrix according to the B-point pixels in the even-numbered rows of pixels.
  • the noise reduction operation circuit when the noise reduction operation is performed on the second pixel, the noise reduction operation circuit further includes: a pixel pair summation subcircuit and a pixel pair difference subcircuit,
  • the pixel pair summation sub-circuit is configured to sum the pixel pairs adjacent to the row in the pixel matrix output by the pixel matrix construction circuit to obtain a second pixel matrix, and input it to the absolute value sub-circuit;
  • the pixel pair difference sub-circuit is configured to obtain the difference of pixel pairs adjacent to rows in the pixel matrix output by the pixel matrix construction circuit to obtain a third pixel matrix, and input it to the absolute value sub-circuit.
  • the noise reduction operation circuit further includes: a second averaging subcircuit configured to average the noise-reduced sum and difference outputted by the first averaging subcircuit to obtain the The second pixel of the center window after noise reduction.
  • the embodiment of the present application further provides a multi-exposure image processing chip, wherein the chip includes any noise reduction circuit provided in the embodiment of the present application.
  • the embodiment of the present application further provides an electronic device, the electronic device comprising: an image acquisition device and any one of the multi-exposure image processing chips provided in the embodiment of the present application.
  • the embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored, wherein when the computer program is executed by a processor, the steps of the multi-exposure image processing method are implemented.
  • the embodiment of the present application also provides a computer program product, including a plurality of instructions, and when the instructions are executed by a computing device, the steps of the aforementioned method are implemented.
  • the size of the pixel window used to perform noise reduction processing on at least two exposure images is respectively determined according to the exposure time, and the at least two exposure images are separately determined according to the determined size of the pixel window.
  • FIG. 1 is a schematic flow chart of the multi-exposure image processing method in the embodiment of the present application. As shown in FIG. 1 , the method may specifically include:
  • Step 101 Obtain at least two exposure images of the target scene captured at different exposure times
  • Step 102 Based on the exposure time of the at least two exposure images, respectively determine the size of the pixel window for performing noise reduction processing on the at least two exposure images;
  • the size of the pixel window includes a width and a height
  • determining the size of the pixel window for performing noise reduction processing on the at least two exposure images based on the exposure time of the at least two exposure images respectively includes: Making at least one of the width and height of the pixel window proportional to the exposure time.
  • the size of the pixel window for the first exposure image is configured as the first width*the first height.
  • the pixel window size for the second exposure image is configured as second width*second height.
  • Step 103 Perform noise reduction processing on the at least two exposure images based on the determined size of the pixel window;
  • performing noise reduction processing on the at least two exposure images based on the determined size of the pixel window includes: performing noise reduction processing on pixels in the central window in the pixel window based on the determined size of the pixel window .
  • the size of the central window is 2 (width) x 2 (height)
  • the central window includes 4 pixel points RGrGbB, which are the points for which the noise reduction operation needs to be performed in the current pixel window.
  • the method also includes: the four Bayer format pixel points in the central window are denoised, and the pixel window is moved by two pixels in the horizontal direction to obtain a new pixel window; for the new pixel window Carry out noise reduction until all pixels of the exposed image are denoised.
  • the embodiment of the present application configures pixel windows of different sizes for different exposure images.
  • the image uses a larger-sized pixel window for noise reduction operations, and for images with less noise interference, a smaller-sized pixel window is used for noise reduction operations, and different processing resources are allocated for different exposure images.
  • a smaller-sized pixel window is used for noise reduction operations, and different processing resources are allocated for different exposure images.
  • Step 104 Perform fusion processing on the at least two exposure images after noise reduction processing, so as to obtain a high dynamic range image of the target scene.
  • the at least two exposure images include a first exposure image captured at a first exposure time and a second exposure image captured at a second exposure time
  • the method further includes: acquiring the The third exposure image captured at the third exposure time of the target scene, wherein the third exposure time is greater than the first exposure time and greater than the second exposure time; the noise reduction processed The first exposure image, the second exposure image after noise reduction processing, and the third exposure image without noise reduction processing are fused to obtain a high dynamic range image of the target scene.
  • noise reduction operation can be performed on each exposure image, and then the exposure images after noise reduction operation can be fused; or a part of the exposure images in the multi-exposure images can be reduced.
  • Noise reduction operation is performed on the other part of the exposure image without noise reduction operation, and then the exposure image after noise reduction processing and the exposure image without noise reduction processing are fused.
  • the noise reduction method provided by the embodiment of the present application can be used to perform noise reduction operations on the short-exposure image and the medium-exposure image, and not perform noise reduction operations on the long-exposure image, and directly fuse them with the images after the noise reduction operation of the other two images, Get an HDR image.
  • the size of the pixel window used to perform noise reduction processing on at least two exposure images is respectively determined according to the exposure time, and the at least two exposure images are processed according to the determined size of the pixel window.
  • Exposed images are subjected to noise reduction processing to meet the noise reduction processing requirements of different exposure images and effectively suppress image noise.
  • it saves processing resources, reduces noise reduction costs, and optimizes the HDR noise reduction solution.
  • the noise reduction operation is further illustrated.
  • the at least two images based on the determined pixel window size are Exposed images are subjected to noise reduction processing including:
  • Step 201 For the first pixel of the four Bayer format pixels in the central window of the pixel window, construct a first pixel matrix using all the first pixels in the pixel window;
  • the exposure image is a Bayer format image.
  • Bayer format is a pixel arrangement method named after the inventor Bayer.
  • an image is divided into many 2x2 blocks, and each 2x2 block has a blue block (Blue, which can be omitted and written as B), and a red block.
  • a block (Red, which can be omitted and written as R) and two green blocks (Green, which can be omitted and written as G), of which the two green blocks can only be in the diagonal position.
  • Gb green block in the same row as the blue block
  • Gr the green block in the same row as the red block
  • Gb and Gr can be considered as different colors.
  • All possible Bayer formats in a 2x2 block have four modes: RGrGbB, BGbGrR, GrRBGb, and GbBRGr.
  • the center window of this application is a 2x2 block at the center of the pixel window.
  • the four Bayer format pixels are illustrated with RGrGbB as an example, and the other three formats are also applicable.
  • the first pixel is a part of pixels in the central window.
  • the first pixel is a green point Gr or a green point Gb among the four Bayer format pixel points.
  • the denoising operation of the middle exposure image of the HDR camera is a Bayer format window of 14 (width) x 10 (height), and the four Bayer format pixels of the central window are RGrGbB It is the point where noise reduction needs to be done in the current pixel window, and other non-central points can be used as noise reduction reference points to participate in noise reduction operations.
  • the noise reduction operation is performed on the green point Gr in the central window, it is performed with the first pixel matrix composed of 7 (width) ⁇ 5 (height) Gr points centered on Gr, as shown in FIG. 4 .
  • the noise reduction operation is performed on the green point Gb of the central window, it is performed with the first pixel matrix composed of 7 (width) ⁇ 5 (height) Gb points centered on Gb, as shown in FIG. 5 .
  • Step 202 Perform a denoising operation on the first pixel matrix to obtain the denoised first pixel of the center window;
  • the noise reduction operation may include averaging the points of the pixel matrix as the first pixel point after noise reduction, or in order to ensure the noise reduction effect, filtering and averaging the points of the pixel matrix as the first pixel point after noise reduction pixel.
  • the noise reduction operation includes: first taking the absolute value of each pixel in the pixel matrix, and then taking the square root to obtain the square root result sqrt_abs_cen of the center point of the pixel matrix and the square root result of the non-central point of the pixel matrix sqrt_abs_ref; subtract the central point sqrt_abs_cen from each non-central point sqrt_abs_ref and take the absolute value to obtain the difference ABSDIFF; determine the target non-central points and the number of target non-central points whose ABSDIFF is smaller than the first threshold; perform all target non-central points After the accumulation, divide by the number of target non-central points to obtain the noise reduction operation result.
  • the first threshold may be the square root result of the center point sqrt_abs_cen, or the threshold may be obtained after calibration of sqrt_abs_cen according to configuration parameters.
  • the method further includes: obtaining configured weights and offsets; multiplying the square root result sqrt_abs_cen of the center point by the weight to obtain a product; combining the product with the offset amount to get the first threshold.
  • Step 1 First take the absolute value of each Gr point in the 7x5 pixel matrix, and then take the square root.
  • the square root result of the marked center Gr point is sqrt_abs_cenGr
  • the square root result of other Gr points is sqrt_abs_Gr;
  • Step 2 Calculate a threshold thresh_cenGr based on the sqrt_abs_cenGr value of the central Gr point. This threshold will participate in subsequent calculations.
  • the value of thresh_cenGr is equal to sqrt_abs_cenGr*reg_weight+reg_offset, where reg_weight and reg_offset are the parameters configured in the register;
  • Step 3 Subtract the sqrt_abs_cenGr of the center Gr point from the 34 sqrt_abs_Gr values to obtain 34 differences, and then take the absolute value of the 34 differences.
  • the calculation result is named after ABSDIFF, and we get a total of 34 ABSDIFF values;
  • Step 4 Compare the 34 ABSDIFF values with the threshold thresh_cenGr, pick out the points whose ABSDIFF values are smaller than the threshold thresh_cenGr, and accumulate these points to obtain a cumulative sum.
  • Step 5 Accumulate and divide by the number of pixels involved in accumulation to obtain the average value of pixels involved in accumulation, and use this average value to replace the central Gr point to complete noise reduction.
  • Step 1 First take the absolute value of each Gb point in the 7x5 pixel matrix, and then take the square root.
  • the square root result of the marked center Gb point is sqrt_abs_cenGb
  • the square root result of other Gb points is sqrt_abs_Gb;
  • Step 2 Calculate a threshold thresh_cenGb based on the sqrt_abs_cenGb value of the central Gb point, which will participate in subsequent calculations.
  • the value of thresh_cenGb is equal to sqrt_abs_cenGb*reg_weight+reg_offset, where reg_weight and reg_offset are the parameters configured in the register;
  • Step 3 Subtract the sqrt_abs_cenGb of the central Gb point from the 34 sqrt_abs_Gb values to obtain 34 differences, and then take the absolute value of the 34 differences.
  • the calculation result is named after ABSDIFF, and we get a total of 34 ABSDIFF values;
  • Step 4 Compare the 34 ABSDIFF values with the threshold thresh_cenGb, pick out the points whose ABSDIFF values are smaller than the threshold thresh_cenGb, and accumulate these points to obtain a cumulative sum.
  • Step 5 Accumulate and divide by the number of pixels involved in accumulation to obtain the average value of pixels involved in accumulation, and use this average value to replace the central Gb point to complete noise reduction.
  • Step 203 For the second pixel in the four Bayer format pixels in the center window of the pixel window, use all the second pixels in the pixel window and their adjacent pixels in the row to form a pixel pair;
  • the second pixel is another part of pixels in the central window.
  • the second pixel point is a red point R or a blue point B among the four Bayer format pixel points.
  • the noise reduction of the red point among the four points in the center of the pixel window is based on all the pixels in the row where the red point is located, and there are 14 (width) x 5 (height) points in total. As shown in Figure 6, adjacent rows can be R and Gr are regarded as a pair, and there are a total of 7x5 such "pixel pairs".
  • the noise reduction of the blue point among the four points in the center of the pixel window is based on all the pixels in the row where the blue point is located. There are 14 (width) x 5 (height) points in total. As shown in Figure 7, the row phase Adjacent B and Gb are regarded as a pair, and there are a total of 7x5 such "pixel pairs".
  • Step 204 Summing each pixel pair, constructing a second pixel matrix using the sum value; calculating the difference for each pixel pair, constructing a third pixel matrix using the difference value;
  • Step 205 performing the noise reduction operation on the second pixel matrix and the third pixel matrix to obtain a sum and a difference after noise reduction;
  • the noise reduction operation includes: first taking the absolute value of each pixel in the pixel matrix, and then taking the square root to obtain the square root result sqrt_abs_cen of the center point of the pixel matrix and the square root result of the non-central point of the pixel matrix sqrt_abs_ref; subtract the central point sqrt_abs_cen from each non-central point sqrt_abs_ref and take the absolute value to obtain the difference ABSDIFF; determine the target non-central points and the number of target non-central points whose ABSDIFF is smaller than the first threshold; perform all target non-central points After the accumulation, divide by the number of target non-central points to obtain the noise reduction operation result.
  • Step 1 Since R and Gr always appear in pairs, sum up 35 pairs of R and Gr respectively to obtain 35 (R+Gr) values. At this time, (R+Gr) is regarded as a pixel, then there are 35 Such pixels, which are exactly the same as the denoising pixel matrix of Gr and Gb, use the same denoising operation to obtain (R+Gr) after denoising;
  • Step 2 Since R and Gr always appear in pairs, calculate the difference between 35 pairs of R and Gr to obtain 35 (R-Gr) values.
  • (R-Gr) is regarded as a pixel, so there are 35 Such a pixel, which is exactly the same as the denoising pixel matrix of Gr and Gb, adopts the same denoising operation to obtain (R-Gr) after denoising;
  • Step 3 Average the two values after denoising (R+Gr) obtained in step 1 and after denoising (R-Gr) obtained in step 2, and replace the central one with the calculation result at this time Point R completes the noise reduction.
  • Step 1 Since B and Gb always appear in pairs, sum up 35 pairs of B and Gb respectively to obtain 35 (B+Gb) values. At this time, (B+Gb) is regarded as a pixel, then there are 35 Such a pixel, which is exactly the same as the color matrix of Gr and Gb noise reduction, using the same algorithm, can get the noise reduction (B+Gb);
  • Step 2 Since B and Gb always appear in pairs, calculate the difference between 35 pairs of B and Gb to obtain 35 (B-Gb) values. At this time, (B-Gb) is regarded as a pixel, so there are 35 in total Such a pixel, which is exactly the same as the color matrix of Gr and Gb noise reduction, using the same algorithm, can get the noise reduction (B-Gb);
  • Step 3 Average the noise-reduced (B+Gb) obtained in step 1 and the noise-reduced (B-Gb) obtained in step 2 again, and replace the center point B with the calculation result at this time , to complete the noise reduction.
  • noise reduction processing of the four points may be performed simultaneously, or may be performed sequentially in a certain order.
  • Step 206 Average the sum and difference after noise reduction to obtain the second pixel point after noise reduction in the center window.
  • the method further includes: the noise reduction of the four Bayer format pixels in the central window is completed, and the pixel window is moved by two pixels in the horizontal direction to obtain a new pixel window;
  • the pixel window of the exposure image is denoised until all the pixels of the exposure image are denoised.
  • the noise reduction of the four rightmost points in the horizontal direction is completed, and the 14x10 pixel window is shifted down by two pixels and returns to the leftmost side of the image to start a new line of noise reduction. noise.
  • the noise reduction of all pixels of the entire frame of image can be completed.
  • the size of the pixel window may be 6 (width) ⁇ 10 (height).
  • the size of the pixel window may be 14 (width) ⁇ 10 (height).
  • the denoising operation of the short-exposure image of the HDR camera the pixel window is a 6 (width) x 10 (height) Bayer format window, and the four Bayer format pixels of the central window are RGrGbB It is the point where noise reduction needs to be done in the current pixel window, and other non-central points can be used as noise reduction reference points to participate in noise reduction operations.
  • the noise reduction of the green point Gr among the four points in the center of the pixel window of the short-exposure image is performed with a pixel matrix composed of 3 (width) ⁇ 5 (height) Gr points centered on Gr, as shown in Figure 9 .
  • the denoising operation steps of the Gr point 3x5 pixel matrix please refer to the denoising operation of the Gr point in the middle exposure image.
  • the noise reduction of the green point Gb is carried out with a pixel matrix composed of 3 (width) x 5 (height) Gr points centered on Gb, as shown in Fig. 10 .
  • a pixel matrix composed of 3 (width) x 5 (height) Gr points centered on Gb, as shown in Fig. 10 .
  • the noise reduction of the red point among the four points in the center of the pixel window of the short-exposure image is based on all the pixels in the row where the red point is located. There are 6 (width) x 5 (height) points in total. As shown in Figure 11, the row Adjacent R and Gr are regarded as a pair, and there are 3x5 such "pixel pairs" in total. For the steps of the noise reduction operation of the 3x5 pixel matrix, please refer to the noise reduction operation of the R point in the medium exposure image.
  • the noise reduction of the blue point among the four points in the center of the pixel window of the short-exposure image is based on all the pixels in the row where the blue point is located, and there are 6 (width) x 5 (height) points in total, as shown in Figure 12, which can be Consider the B and Gb adjacent to each other as a pair, and there are 3x5 such "pixel pairs" in total.
  • the noise reduction operation of the 3x5 pixel matrix please refer to the noise reduction operation of point B in the medium exposure image.
  • noise reduction is done for the four points in the center of the short-exposure pixel window.
  • noise reduction Repeat the above operations until the noise reduction of the four rightmost points in the horizontal direction is completed, the 6x10 pixel window is shifted down by two pixels and returns to the leftmost side of the image, and a new line of noise reduction is started.
  • the size of the pixel window used to perform noise reduction processing on at least two exposure images is respectively determined according to the exposure time, and the at least two exposure images are processed according to the determined size of the pixel window.
  • Exposed images are subjected to noise reduction processing to meet the noise reduction processing requirements of different exposure images and effectively suppress image noise.
  • it saves processing resources, reduces noise reduction costs, and optimizes the HDR noise reduction solution.
  • the embodiment of the present application also provides a multi-exposure image processing device, as shown in FIG. 13 , the multi-exposure image processing device includes:
  • the acquisition module 131 is configured to acquire at least two exposure images of the target scene captured at different exposure times;
  • the noise reduction module 132 is configured to, based on the exposure time of the at least two exposure images, respectively determine the size of the pixel window for performing noise reduction processing on the at least two exposure images; performing noise reduction processing on the at least two exposure images;
  • the fusion module 133 is configured to perform fusion processing on the at least two exposure images after noise reduction processing, so as to obtain a high dynamic range image of the target scene.
  • the noise reduction module 132 includes:
  • a pixel window construction circuit 141 configured to load the pixel window of the exposure image according to the size of the pixel window
  • An odd-numbered row extraction circuit 142 configured to extract odd-numbered row pixels from the pixel window loaded by the pixel window construction circuit 141;
  • An even-numbered row extraction circuit 143 configured to extract even-numbered row pixels from the pixel window loaded by the pixel window construction circuit 141;
  • a pixel matrix construction circuit 144 configured to construct a pixel matrix according to the pixels in the odd rows or the pixels in the even rows;
  • the noise reduction operation circuit 145 is configured to perform a noise reduction operation on the pixel matrix to output noise-reduced pixels.
  • the pixel matrix construction submodule is configured to use all the first pixels in the pixel window to construct a first pixel matrix for the first pixel among the four Bayer format pixels in the central window of the pixel window;
  • the denoising operation sub-module is configured to perform the denoising operation on the first pixel matrix to obtain the denoised first pixel of the central window;
  • the pixel matrix construction sub-module is also used for, for the second pixel in the four Bayer format pixels of the central window of the pixel window, using all the second pixels in the pixel window and the adjacent pixels of the row.
  • the points form a pixel pair; sum each pixel pair, and use the sum value to construct the second pixel matrix; calculate the difference for each pixel pair, and use the difference value to construct the third pixel matrix;
  • the noise reduction operation sub-module is configured to perform the noise reduction operation on the second pixel matrix and the third pixel matrix to obtain the sum and difference after noise reduction; average the sum and difference after noise reduction , to obtain the second pixel of the center window after noise reduction.
  • the first pixel is the green point Gr or the green point Gb in the four Bayer format pixels
  • the second pixel is the green point Gb in the four Bayer format pixels Red point R point or blue point B point.
  • the noise reduction operation sub-module is specifically used to take the absolute value of each pixel in the pixel matrix first, and then take the square root to obtain the square root result sqrt_abs_cen of the center point of the pixel matrix and the root of the non-central point of the pixel matrix Square the result sqrt_abs_ref; subtract the central point sqrt_abs_cen from each non-central point sqrt_abs_ref and take the absolute value to obtain the difference ABSDIFF; determine the target non-central point and the number of target non-central points whose ABSDIFF is smaller than the first threshold; set all target non-central Points are accumulated and divided by the number of target non-central points to obtain the result of the noise reduction operation.
  • the noise reduction operator module is also used to obtain the configured weight and offset; multiply the square root result sqrt_abs_cen of the center point by the weight to obtain the product; Adding the product and the offset to obtain the first threshold.
  • the noise reduction module 132 is also used to complete the noise reduction of the four Bayer format pixels in the central window, and moves the pixel window by two pixels in the horizontal direction to obtain a new Pixel window: performing noise reduction processing on the new pixel window until the noise reduction of all pixels of the exposure image is completed.
  • the size of the pixel window includes width and height
  • the noise reduction module 132 is further configured to make at least one of the width and height of the pixel window proportional to the exposure time .
  • the at least two exposure images include a first exposure image captured at a first exposure time and a second exposure image captured at a second exposure time
  • the acquisition module 131 It is configured to acquire a third exposure image captured by the target scene at a third exposure time, wherein the third exposure time is greater than the first exposure time and greater than the second exposure time;
  • the fusion module 133 is configured to fuse the first exposure image after noise reduction processing, the second exposure image after noise reduction processing, and the third exposure image without noise reduction processing, to obtain a high dynamic range image of the target scene.
  • the embodiment of the present application also provides a noise reduction circuit.
  • the noise reduction circuit includes:
  • the pixel window construction circuit 141 is configured to load the pixel window of the exposure image according to the size of the pixel window for performing noise reduction processing on the exposure image;
  • An odd-numbered row extraction circuit 142 configured to extract odd-numbered row pixels from the pixel window loaded by the pixel window construction circuit 141;
  • An even-numbered row extraction circuit 143 configured to extract even-numbered row pixels from the pixel window loaded by the pixel window construction circuit 141;
  • a pixel matrix construction circuit 144 configured to construct a pixel matrix according to the pixels in the odd rows or the pixels in the even rows;
  • the noise reduction operation circuit 145 is configured to perform a noise reduction operation on the pixel matrix to output noise-reduced pixels.
  • the noise reduction circuit further includes a line buffer 146 configured to buffer pixels of H-1 line pixel windows, where H is the height of the pixel window, and when the Hth line of pixels is obtained, the pixel window construction circuit 141 loads a complete pixel window.
  • the pixel matrix construction circuit 144 includes: a Gr point pixel matrix construction circuit, an R point pixel matrix construction circuit, a B point pixel matrix construction circuit, and a Gb point pixel matrix construction circuit.
  • the R points and the Gr points are located in odd-numbered rows, and the Gb points and B points are located in even-numbered rows.
  • Gr point pixel matrix namely the first pixel matrix
  • R point pixel matrix namely the first pixel matrix
  • the B-point pixel matrix construction circuit and the Gb-point pixel matrix construction circuit construct the Gb-point pixel matrix (ie, the first pixel matrix) and the B-point pixel matrix according to the even-numbered row pixels output by the even-numbered row extraction circuit 143 .
  • the noise reduction operation circuit 145, the Gr point noise reduction operation circuit, the R point noise reduction operation circuit, the B point noise reduction operation circuit and the Gb point noise reduction operation circuit respectively perform noise reduction operations on the four points.
  • the noise reduction circuit further includes an output circuit 147 configured to arrange and output the output order of the four points after noise reduction.
  • the noise reduction circuit includes: 9-line buffer and 14x10 pixel window construction circuit, the 9-line buffer is used to buffer the first 9 lines of pixels, when the 10th line of pixels is received, The pixel window building circuit builds a 14x10 pixel window.
  • the noise reduction circuit includes: 9 line buffers and a 6x10 pixel window construction circuit.
  • the noise reduction operation circuit 145 includes:
  • Obtaining an absolute value sub-circuit 151 configured to obtain an absolute value for each point in the pixel matrix
  • the square root sub-circuit 152 is configured to take the square root of the absolute value of each point in the pixel matrix to obtain the square root result sqrt_abs_cen of the center point of the pixel matrix and the square root result sqrt_abs_ref of the non-central point of the pixel matrix;
  • the comparison sub-circuit 154 is configured to compare the ABSDIFF with a preset threshold, and determine the target non-central points and the number of target non-central points whose ABSDIFF is smaller than the first threshold;
  • the first averaging sub-circuit 155 is configured to accumulate all target non-central points and divide by the number of target non-central points to obtain denoised pixels.
  • the noise reduction operation circuit 145 further includes: a threshold configuration subcircuit 156 configured to obtain the weight and offset of the register configuration; multiply the square root result sqrt_abs_cen of the center point by the weight , to obtain a product; add the product and the offset to obtain the first threshold.
  • the first threshold is equal to sqrt_abs_cenGr*reg_weight+reg_offset, where reg_weight and reg_offset are the weight and offset configured in the register.
  • the noise reduction operation circuit 145 when performing noise reduction operations on the pixel matrix at point R and the pixel matrix at point B, the noise reduction operation circuit 145 also includes: a pixel pair summation subcircuit 157 and a pixel pair difference subcircuit 158 ,
  • the pixel pair summation sub-circuit 157 is configured to sum the pixel pairs adjacent to the row in the pixel matrix output by the pixel matrix construction circuit to obtain a second pixel matrix, and input it to the absolute value sub-circuit;
  • the pixel pair subtraction sub-circuit 158 is configured to calculate the difference of pixel pairs adjacent to rows in the pixel matrix output by the pixel matrix construction circuit to obtain a third pixel matrix, and input it to the absolute value acquisition sub-circuit.
  • the middle is the same as in Fig. 15, except that the second averaging subcircuit 159 is added at the end, configured to average the sum and difference of the noise-reduced output of the first averaging subcircuit to obtain the reduced value of the center window.
  • the second pixel after noise is the same as in Fig. 15, except that the second averaging subcircuit 159 is added at the end, configured to average the sum and difference of the noise-reduced output of the first averaging subcircuit to obtain the reduced value of the center window.
  • the second pixel after noise is the middle.
  • the Gr noise reduction operation circuit and the Gb noise reduction operation circuit may be the noise reduction operation circuit shown in FIG. 15
  • the R noise reduction operation circuit and the B noise reduction operation circuit may be the noise reduction operation circuit shown in FIG. 16 .
  • the noise reduction module in the multi-exposure image processing device can be realized by the above noise reduction circuit.
  • the present application also provides a multi-exposure image processing chip, as shown in FIG. 17 , the chip 170 includes any noise reduction circuit 1701 in the above-mentioned embodiments.
  • the multi-exposure image processing chip 170 may also include an input interface 1702 . Communicate with other devices or chips through the input interface 1702 , specifically, information or data sent by other devices or chips can be obtained.
  • the multi-exposure image processing chip 170 may also include an output interface 1703 . Communicate with other devices or chips through the output interface 1703 , specifically, output information or data to other devices or chips.
  • the multi-exposure image processing chip 170 may be applied to the electronic device in the embodiment of the present application.
  • the chip mentioned in the embodiment of the present application may also be called a system-on-chip, a system-on-chip, a system-on-a-chip, or a system-on-a-chip.
  • the above-mentioned processor can be application specific integrated circuit (ASIC, Application Specific Integrated Circuit), digital signal processing device (DSPD, Digital Signal Processing Device), programmable logic device (PLD, Programmable Logic Device), on-site At least one of a programmable gate array (Field-Programmable Gate Array, FPGA), a controller, a microcontroller, and a microprocessor.
  • ASIC Application Specific Integrated Circuit
  • DSPD Digital Signal Processing Device
  • PLD Programmable Logic Device
  • FPGA Field-Programmable Gate Array
  • controller a microcontroller
  • microprocessor programmable gate array
  • the electronic device configured to implement the above processor function may also be other, which is not specifically limited in this embodiment of the present application.
  • memory can be volatile memory (volatile memory), such as random access memory (RAM, Random-Access Memory); Or non-volatile memory (non-volatile memory), such as read-only memory (ROM, Read-Only Memory), flash memory (flash memory), hard disk (HDD, Hard Disk Drive) or solid-state drive (SSD, Solid-State Drive); or a combination of the above types of memory, and provide instructions and data to the processor.
  • volatile memory such as random access memory (RAM, Random-Access Memory
  • non-volatile memory such as read-only memory (ROM, Read-Only Memory), flash memory (flash memory), hard disk (HDD, Hard Disk Drive) or solid-state drive (SSD, Solid-State Drive); or a combination of the above types of memory, and provide instructions and data to the processor.
  • the embodiment of the present application also provides an electronic device.
  • the electronic device described in the present application has a shooting function and can take exposure images with different exposure times.
  • the electronic device can include mobile phones, tablet computers, notebook computers, palmtop computers, personal digital assistants ( Personal Digital Assistant, PDA), portable media player (Portable Media Player, PMP), wearable devices, cameras, smart cars, etc.
  • PDA Personal Digital Assistant
  • PMP portable media player
  • wearable devices cameras, smart cars, etc.
  • the electronic device 180 includes: an image acquisition device 1801 and the aforementioned multi-exposure image processing chip 1802 .
  • the image acquisition device 1801 is configured to acquire at least two exposure images captured at least two exposure times of the target scene;
  • the multi-exposure image processing chip 1802 is any one of the multi-exposure image processing chips in the above embodiments, and is configured to perform noise reduction processing on different exposure images using different noise reduction processing strategies.
  • the multi-exposure image processing chip 1802 is also configured to fuse the noise-reduced exposure images.
  • bus system 1803 various components in the electronic device 180 are coupled together through a bus system 1803 .
  • the bus system 1803 is used to realize connection and communication between these components.
  • the bus system 1803 also includes a power bus, a control bus and a status signal bus.
  • the various buses are labeled as bus system 1803 in FIG. 18 for clarity of illustration.
  • the embodiment of the present application further provides a computer-readable storage medium, such as a memory including a computer program, and the computer program can be executed by a processor to complete the steps of the aforementioned method.
  • a computer-readable storage medium such as a memory including a computer program
  • the embodiment of the present application also provides a computer program product, including a plurality of instructions, and when the instructions are executed by a computing device, the steps of the multi-exposure image processing method in the embodiment of the present application are implemented.
  • the computer program product may be applied to the processor in the embodiments of the present application, and the computer program instructions cause the computer to execute the corresponding processes implemented by the processor in the various methods of the embodiments of the present application.
  • the Let me repeat for the sake of brevity, the Let me repeat.
  • the embodiment of the present application also provides a computer program.
  • the computer program can be applied to the processor in the embodiment of the present application, and when the computer program is run on the computer, the computer executes the corresponding process implemented by the processor in each method of the embodiment of the present application, for the sake of brevity , which will not be repeated here.
  • first, second, third, etc. may be used in this application to describe various information, the information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another and are not necessarily used to describe a specific order or sequence.
  • first information may also be called second information, and similarly, second information may also be called first information.
  • the units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units; Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.
  • each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can be used as a single unit, or two or more units can be integrated into one unit; the above-mentioned integration
  • the unit can be realized in the form of hardware or in the form of hardware plus software functional unit.
  • Embodiments of the present application provide a multi-exposure image processing method, device, and noise reduction circuit.
  • the method determines according to the exposure time to perform noise reduction processing on at least two exposure images.
  • the size of the pixel window according to the size of the determined pixel window, perform noise reduction processing on at least two exposure images respectively, so as to meet the noise reduction processing requirements of different exposure images and effectively suppress image noise.
  • the noise reduction effect saving Process resources, reduce noise reduction costs, and optimize HDR noise reduction solutions.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)

Abstract

本申请公开了一种多曝光图像处理方法、装置及降噪电路，该方法对HDR摄影的多曝光图像进行处理时，根据曝光时间分别确定用于对至少两幅曝光图像执行降噪处理的像素窗口的尺寸，根据确定的像素窗口的尺寸对至少两幅曝光图像分别执行降噪处理，从而满足不同曝光图像的降噪处理需求有效抑制图像噪声，在保证降噪效果的基础上，节约处理资源，降低降噪成本，优化HDR的降噪方案。

Description

多曝光图像处理方法、装置及降噪电路
相关申请的交叉引用
本申请基于申请号为202111129165.5、申请日为2021年09月26日、发明创造名称为“一种多曝光图像处理方法、装置及降噪电路”的在先中国专利申请提出，并要求该在先中国专利申请的优先权，该在先中国专利申请的全部内容在此以全文引入的方式引入本申请作为参考。
技术领域
本申请的实施例涉及图像处理技术，尤其涉及一种多曝光图像处理方法、装置及降噪电路。
背景技术
高动态范围(High Dynamic Range，HDR)摄影的实现有多种方法，目前较为主流的方法是数字重叠(Digital Overlap，DOL)，其具体做法是对同一场景分别做长曝光、中曝光和短曝光，采集到三张图像，然后把这三张图像做融合，最后得到一张高动态范围的图像。在对三张图像做融合时，需要考虑三张图像中的噪声对融合结果的影响，一般需要对图像做降噪处理。具体实施降噪时，可以选择在融合之前降噪，也可以选择在融合之后降噪。因此，多曝光图像的降噪问题是一个需要长期研究并解决的问题。
发明内容
本申请实施例期望提供一种多曝光图像处理方法、装置及降噪电路。
本申请的技术方案是这样实现的：
第一方面，提供了一种多曝光图像处理方法，包括：
获取目标场景在不同曝光时间下拍摄到的至少两幅曝光图像；
基于所述至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸；
基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理；
对经过降噪处理后的所述至少两幅曝光图像进行融合处理，以获得所述目标场景的高动态范围图像。
第二方面，提供了一种多曝光图像处理装置，包括：
获取模块，配置为获取目标场景在不同曝光时间下拍摄到的至少两幅曝光图像；
降噪模块，配置为基于所述至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸；基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理；
融合模块，配置为对经过降噪处理后的所述至少两幅曝光图像进行融合处理，以获得所述目标场景的高动态范围图像。
第三方面，提供了一种降噪电路，所述降噪电路包括：
像素窗口构建电路，配置为根据曝光图像执行降噪处理的像素窗口的尺寸加载所述 曝光图像的像素窗口；
奇数行提取电路，配置为从所述像素窗口构建电路加载的像素窗口中提取奇数行像素；
偶数行提取电路，配置为从所述像素窗口构建电路加载的像素窗口中提取偶数行像素；
像素矩阵构建电路，配置为根据所述奇数行像素或所述偶数行像素构建像素矩阵；
降噪运算电路，配置为对所述像素矩阵进行降噪运算输出降噪后的像素点。
第四方面，提供了一种多曝光图像处理芯片，所述芯片包括前述第三方面所述的降噪电路。
第五方面，提供了一种电子设备，所述电子设备包括：图像采集装置以及前述多曝光图像处理芯片。
第六方面，提供了一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现前述方法的步骤。
第七方面，提供了一种计算机程序产品，包括多个指令，所述指令被计算设备执行时实现前述方法的步骤。
附图说明
图1为本申请实施例中多曝光图像处理方法的流程示意图；
图2为本申请实施例中降噪处理方法的流程示意图；
图3为本申请实施例中像素窗口的第一尺寸示意图；
图4为本申请实施例中Gr点像素矩阵的第一示意图；
图5为本申请实施例中Gb点像素矩阵的第一示意图；
图6为本申请实施例中R点像素矩阵的第一示意图；
图7为本申请实施例中B点像素矩阵的第一示意图；
图8为本申请实施例中像素窗口的第二尺寸示意图；
图9为本申请实施例中Gr点像素矩阵的第二示意图；
图10为本申请实施例中Gb点像素矩阵的第二示意图；
图11为本申请实施例中R点像素矩阵的第二示意图；
图12为本申请实施例中B点像素矩阵的第二示意图；
图13为本申请实施例中多曝光图像处理装置的组成结构示意图；
图14为本申请实施例中降噪电路的组成结构示意图；
图15为本申请实施例中降噪运算电路的第一组成结构示意图；
图16为本申请实施例中降噪运算电路的第二组成结构示意图；
图17为本申请实施例中多曝光图像处理芯片的组成结构示意图；
图18为本申请实施例中电子设备的组成结构示意图。
具体实施方式
本申请实施例提供了一种多曝光图像处理方法，包括：
获取目标场景在不同曝光时间下拍摄到的至少两幅曝光图像；
基于所述至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸；
基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理；
对经过降噪处理后的所述至少两幅曝光图像进行融合处理，以获得所述目标场景的 高动态范围图像。
在一些实施例中，所述至少两幅曝光图像为Bayer格式图像，所述基于所述像素窗口对所述至少两幅曝光图像分别执行降噪处理包括：
对于所述像素窗口的中心窗口的四个Bayer格式像素点中的第一像素点，利用所述像素窗口中的全部第一像素点构建第一像素矩阵；
对所述第一像素矩阵进行降噪运算，得到所述中心窗口的降噪后的第一像素点；
对于所述像素窗口的中心窗口的四个Bayer格式像素点中的第二像素点，利用所述像素窗口中的所有第二像素点及其行相邻的像素点组成一个像素对；
对每个像素对求和，利用和值构建第二像素矩阵；
对每个像素对求差，利用差值构建第三像素矩阵；
对所述第二像素矩阵和第三像素矩阵进行所述降噪运算，得到降噪后的和值和差值；
对降噪后的和值和差值求平均，得到所述中心窗口的降噪后的第二像素点。
在一些实施例中，所述第一像素点为所述四个Bayer格式像素点中的绿色点Gr点或者绿色点Gb点，
所述第二像素点为所述四个Bayer格式像素点中的红色点R点或者蓝色点B点。
在一些实施例中，所述降噪运算包括：
对像素矩阵中每个像素点先取绝对值，再取开平方，得到像素矩阵中心点的开平方结果sqrt_abs_cen和像素矩阵非中心点的开平方结果sqrt_abs_ref；
将每个非中心点sqrt_abs_ref减去中心点sqrt_abs_cen并取绝对值，得到差值ABSDIFF；
确定ABSDIFF小于第一阈值的目标非中心点以及目标非中心点个数；
将全部目标非中心点进行累加后除以所述目标非中心点个数，得到降噪运算结果。
在一些实施例中，所述方法还包括：
获取配置的权重和偏移量；
将所述中心点的开平方结果sqrt_abs_cen和所述权重相乘，得到乘积；
将所述乘积和所述偏移量相加，得到所述第一阈值。
在一些实施例中，所述方法还包括：
所述中心窗口中的四个Bayer格式像素点降噪完成，将所述像素窗口沿水平方向移动两个像素，得到新的像素窗口；
对所述新的像素窗口进行降噪处理，直到曝光图像的全部像素点降噪完成。
在一些实施例中，所述像素窗口的尺寸包括宽度和高度，所述基于至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸包括：
使所述像素窗口的宽度和高度中的至少一项与所述曝光时间成正比。
在一些实施例中，所述至少两幅曝光图像包括在第一曝光时间下拍摄到的第一曝光图像、在第二曝光时间下拍摄到的第二曝光图像，所述方法还包括：
获取所述目标场景在第三曝光时间下拍摄到的第三曝光图像，其中所述第三曝光时间大于所述第一曝光时间，且大于所述第二曝光时间；
将经过降噪处理后的所述第一曝光图像，经过降噪处理后的所述第二曝光图像，以及未经过降噪处理的所述第三曝光图像进行融合处理，以得到所述目标场景的高动态范围图像。
本申请实施例还提供了一种多曝光图像处理装置，包括：
获取模块，配置为获取目标场景在不同曝光时间下拍摄到的至少两幅曝光图像；
降噪模块，配置为基于所述至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸；基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理；
融合模块，配置为对经过降噪处理后的所述至少两幅曝光图像进行融合处理，以获得所述目标场景的高动态范围图像。
本申请实施例还提供了一种降噪电路，所述降噪电路包括：
像素窗口构建电路，配置为根据曝光图像执行降噪处理的像素窗口的尺寸加载所述曝光图像的像素窗口；
奇数行提取电路，配置为从所述像素窗口构建电路加载的像素窗口中提取奇数行像素；
偶数行提取电路，配置为从所述像素窗口构建电路加载的像素窗口中提取偶数行像素；
像素矩阵构建电路，配置为根据所述奇数行像素或所述偶数行像素构建像素矩阵；
降噪运算电路，配置为对所述像素矩阵进行降噪运算输出降噪后的像素点。
在一些实施例中，所述降噪运算电路包括：
取绝对值子电路，配置为对所述像素矩阵中每个点取绝对值；
开平方子电路，配置为对所述像素矩阵中每个点的绝对值开平方，得到所述像素矩阵中心点的开平方结果sqrt_abs_cen和像素矩阵非中心点的开平方结果sqrt_abs_ref；
作差取绝对值子电路，配置为将每个非中心点sqrt_abs_ref减去中心点sqrt_abs_cen并取绝对值，得到差值ABSDIFF；
比较子电路，配置为将所述ABSDIFF和预设阈值比较，确定ABSDIFF小于第一阈值的目标非中心点以及目标非中心点个数；
第一求平均子电路，配置为将全部目标非中心点进行累加后除以所述目标非中心点个数，得到降噪后的像素点。
在一些实施例中，所述降噪运算电路还包括：
阈值配置子电路，配置为获取寄存器配置的权重和偏移量；将所述中心点的开平方结果sqrt_abs_cen和所述权重相乘，得到乘积；将所述乘积和所述偏移量相加，得到所述第一阈值。
在一些实施例中，所述像素矩阵构建电路包括：Gr点像素矩阵构建电路和Gb点像素矩阵构建电路；
所述Gr点像素矩阵构建电路，配置为根据所述奇数行像素中的Gr点像素构建Gr点像素矩阵；
所述Gb点像素矩阵构建电路，配置为根据所述偶数行像素中的Gb点像素构建Gb点像素矩阵。
在一些实施例中，所述像素矩阵构建电路包括：R点像素矩阵构建电路和B点像素矩阵构建电路；
所述R点像素矩阵构建电路，配置为根据所述奇数行像素中的R点像素构建R点像素矩阵；
所述B点像素矩阵构建电路，配置为根据所述偶数行像素中的B点像素构建B点像素矩阵。
在一些实施例中，对第二像素点进行降噪运算时，所述降噪运算电路还包括：像素对求和子电路和像素对求差子电路，
所述像素对求和子电路，配置为对所述像素矩阵构建电路输出的像素矩阵中行相邻的像素对求和，得到第二像素矩阵，并输入到取绝对值子电路；
所述像素对求差子电路，配置为对所述像素矩阵构建电路输出的像素矩阵中行相邻的像素对求差，得到第三像素矩阵，并输入到所述取绝对值子电路。
在一些实施例中，所述降噪运算电路还包括：第二求平均子电路，配置为对所述第一求平均子电路输出的降噪后的和值和差值求平均，得到所述中心窗口的降噪后的第二像素点。
本申请实施例还提供了一种多曝光图像处理芯片，其中，所述芯片包括本申请实施例提供的任一种降噪电路。
本申请实施例还提供了一种电子设备，所述电子设备包括：图像采集装置以及本申请实施例提供的任一种多曝光图像处理芯片。
本申请实施例还提供了一种计算机可读存储介质，其上存储有计算机程序，其中，所述计算机程序被处理器执行时实现多曝光图像处理方法的步骤。
本申请实施例还提供了一种计算机程序产品，包括多个指令，所述指令被计算设备执行时实现前述方法的步骤。
如此，对HDR摄影的多曝光图像进行处理时，根据曝光时间分别确定用于对至少两幅曝光图像执行降噪处理的像素窗口的尺寸，根据确定的像素窗口的尺寸对至少两幅曝光图像分别执行降噪处理，从而满足不同曝光图像的降噪处理需求有效抑制图像噪声，在保证降噪效果的基础上，节约处理资源，降低降噪成本，优化HDR的降噪方案。
为了能够更加详尽地了解本申请实施例的特点与技术内容，下面结合附图对本申请实施例的实现进行详细阐述，所附附图仅供参考说明之用，并非用来限定本申请实施例。
下面对本申请实施例提供了一种多曝光图像处理方法进行详细的举例说明，图1为本申请实施例中多曝光图像处理方法的流程示意图，如图1所示，该方法具体可以包括：
步骤101：获取目标场景在不同曝光时间下拍摄到的至少两幅曝光图像；
步骤102：基于所述至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸；
示例性的，所述像素窗口的尺寸包括宽度和高度，所述基于至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸包括：使所述像素窗口的宽度和高度中的至少一项与所述曝光时间成正比。
也就是说，曝光时间越长像素窗口的尺寸越大。示例性的，对于第一曝光图像的像素窗口尺寸配置为第一宽度*第一高度。对于第二曝光图像的像素窗口尺寸配置为第二宽度*第二高度。第一曝光时间小于第二曝光时间时，第一宽度小于第二宽度，和/或，第一高度小于第二高度。
步骤103：基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理；
在一些实施例中，基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理包括：基于所确定的像素窗口的尺寸对像素窗口内中心窗口的像素点进行降噪处理。
这里，中心窗口尺寸为2(宽度)x2(高度)，中心窗口内包括4个像素点RGrGbB是当前像素窗口需要进行降噪运算的点。进一步的，该方法还包括：所述中心窗口中的四个Bayer格式像素点降噪完成，将所述像素窗口沿水平方向移动两个像素，得到新的像素窗口；对所述新的像素窗口进行降噪处理，直到曝光图像的全部像素点降噪完成。
在HDR摄影中考虑到不同曝光时间下拍摄的曝光图像的噪声不同，以及不同曝光图像对HDR图像贡献不同，本申请实施例为不同曝光图像配置了不同尺寸的像素窗口，对于噪声干扰较大的图像采用较大尺寸的像素窗口进行降噪运算，对于噪声干扰较小的图像采用较小尺寸的像素窗口进行降噪运算，为不同曝光图像分配不同的处理资源，在 保证降噪效果的基础上，实现的处理资源的合理分配，提高处理资源的利用率，降低降噪成本。
步骤104：对经过降噪处理后的所述至少两幅曝光图像进行融合处理，以获得所述目标场景的高动态范围图像。
在一些实施例中，所述至少两幅曝光图像包括在第一曝光时间下拍摄到的第一曝光图像、在第二曝光时间下拍摄到的第二曝光图像，所述方法还包括：获取所述目标场景在第三曝光时间下拍摄到的第三曝光图像，其中所述第三曝光时间大于所述第一曝光时间，且大于所述第二曝光时间；将经过降噪处理后的所述第一曝光图像，经过降噪处理后的所述第二曝光图像，以及未经过降噪处理的所述第三曝光图像进行融合处理，以得到所述目标场景的高动态范围图像。
也就是说，对于HDR摄影中的多曝光图像，可以对每幅曝光图像均进行降噪运算，再对降噪运算后的曝光图像进行融合处理；或者对多曝光图像中的一部分曝光图像进行降噪运算，另一部分曝光图像不进行降噪运算，再将经过降噪处理后的曝光图像和未经过降噪处理的曝光图像进行融合处理。
示例性的，在HDR摄影中常采用三种曝光时间获取三张曝光图像，短曝光图像、中曝光图像和长曝光图像，分别对应第一曝光时间、第二曝光时间和第三曝光时间，第一曝光时间小于第二曝光时间，第二曝光时间小于第三曝光时间。可以采用本申请实施例提供的降噪方法对短曝光图像和中曝光图像进行降噪运算，对长曝光图像不进行降噪运算，直接与其他两幅图像的降噪运算后的图像进行融合，得到HDR图像。
采用上述技术方案，对HDR摄影的多曝光图像进行处理时，根据曝光时间分别确定用于对至少两幅曝光图像执行降噪处理的像素窗口的尺寸，根据确定的像素窗口的尺寸对至少两幅曝光图像分别执行降噪处理，从而满足不同曝光图像的降噪处理需求有效抑制图像噪声，在保证降噪效果的基础上，节约处理资源，降低降噪成本，优化HDR的降噪方案。
基于上述实施例，以曝光图像为Bayer格式图像为例，对降噪运算进行进一步举例说明，示例性的，如图2所示，所述基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理包括：
步骤201：对于所述像素窗口的中心窗口的四个Bayer格式像素点中的第一像素点，利用所述像素窗口中的全部第一像素点构建第一像素矩阵；
这里，曝光图像是Bayer格式图像。所谓Bayer格式，是以发明人Bayer名字命名的像素排列方法，首先把一张图像划分为很多个2x2块，每个2x2块中有一个蓝色块(Blue，可省略写为B)，一个红色块(Red，可省略写为R)和两个绿色块(Green，可省略写为G)，其中两个绿色块只能处于对角线的位置。一般为了区分两个绿色块，把与蓝色块在同一行的那个绿色块称为Gb，把与红色块在同一行的那个绿色块称为Gr，Gb和Gr可认为是不同的颜色。2x2块中所有可能的Bayer格式有RGrGbB、BGbGrR、GrRBGb和GbBRGr四种模式。本申请中心窗口为像素窗口中心位置的2x2块，四个Bayer格式像素点以RGrGbB为例来进行举例说明，其他三种格式同样适用。
第一像素点为中心窗口内的一部分像素点。示例性的，第一像素点为所述四个Bayer格式像素点中的绿色点Gr点或者绿色点Gb点。
示例性的，如图3所示，对HDR摄像的中曝光图像的降噪运算，以像素窗口是14(宽度)x10(高度)的Bayer格式的窗口，中心窗口的四个Bayer格式像素点RGrGbB是当前像素窗口需要做降噪的点，其他非中心点作可以为降噪参考点参与降噪运算。
对中心窗口的绿色点Gr进行降噪运算时，是以Gr为中心的7(宽度)x5(高度)个Gr点组成的第一像素矩阵进行的，如图4所示。
对中心窗口的绿色点Gb进行降噪运算时，是以Gb为中心的7(宽度)x5(高度)个Gb点组成的第一像素矩阵进行的，如图5所示。
步骤202：对所述第一像素矩阵进行降噪运算，得到所述中心窗口的降噪后的第一像素点；
示例性的，降噪运算可以包括对像素矩阵的点取平均值作为降噪后的第一像素点，或者为了保证降噪效果，对像素矩阵的点过滤后取平均作为降噪后的第一像素点。
在一些实施例中，所述降噪运算包括：对像素矩阵中每个像素点先取绝对值，再取开平方，得到像素矩阵中心点的开平方结果sqrt_abs_cen和像素矩阵非中心点的开平方结果sqrt_abs_ref；将每个非中心点sqrt_abs_ref减去中心点sqrt_abs_cen并取绝对值，得到差值ABSDIFF；确定ABSDIFF小于第一阈值的目标非中心点以及目标非中心点个数；将全部目标非中心点进行累加后除以所述目标非中心点个数，得到降噪运算结果。
示例性的，第一阈值可以为中心点的开平方结果sqrt_abs_cen，或者根据配置参数对sqrt_abs_cen校准后得到阈值。
在一些实施例中，所述方法还包括：获取配置的权重和偏移量；将所述中心点的开平方结果sqrt_abs_cen和所述权重相乘，得到乘积；将所述乘积和所述偏移量相加，得到所述第一阈值。
示例性的，对图4的Gr点做降噪运算的步骤如下：
步骤1：对7x5像素矩阵中的每一个Gr点都先取绝对值，再取开平方，为方便陈述，标记中心Gr点的开平方结果为sqrt_abs_cenGr，其他Gr点的开平方结果为sqrt_abs_Gr；
步骤2：基于中心Gr点的sqrt_abs_cenGr值来计算一个阈值thresh_cenGr，这个阈值会参与后续计算，thresh_cenGr的值等于sqrt_abs_cenGr*reg_weight+reg_offset，其中reg_weight和reg_offset是寄存器配置进来的参数；
步骤3：用34个sqrt_abs_Gr值分别减去中心Gr点的sqrt_abs_cenGr，得到34个差值，然后对这34个差值取绝对值，计算结果以ABSDIFF命名，我们一共得到34个ABSDIFF值；
步骤4：拿34个ABSDIFF值分别与阈值thresh_cenGr做比较，挑出ABSDIFF值小于阈值thresh_cenGr的点，对这些点做累加，得到一个累加和。
步骤5：累加和除以参与累加的像素点的个数，得到参与累加的像素点的平均值，用这个平均值取代中心Gr点，完成降噪。
与Gr点降噪运算过程相同，对图5的Gb点做降噪运算的步骤如下：
步骤1：对7x5像素矩阵中的每一个Gb点都先取绝对值，再取开平方，为方便陈述，标记中心Gb点的开平方结果为sqrt_abs_cenGb，其他Gb点的开平方结果为sqrt_abs_Gb；
步骤2：基于中心Gb点的sqrt_abs_cenGb值来计算一个阈值thresh_cenGb，这个阈值会参与后续计算，thresh_cenGb的值等于sqrt_abs_cenGb*reg_weight+reg_offset，其中reg_weight和reg_offset是寄存器配置进来的参数；
步骤3：用34个sqrt_abs_Gb值分别减去中心Gb点的sqrt_abs_cenGb，得到34个差值，然后对这34个差值取绝对值，计算结果以ABSDIFF命名，我们一共得到34个ABSDIFF值；
步骤4：拿34个ABSDIFF值分别与阈值thresh_cenGb做比较，挑出ABSDIFF值小于阈值thresh_cenGb的点，对这些点做累加，得到一个累加和。
步骤5：累加和除以参与累加的像素点的个数，得到参与累加的像素点的平均值，用这个平均值取代中心Gb点，完成降噪。
步骤203：对于所述像素窗口的中心窗口的四个Bayer格式像素点中的第二像素点，利用所述像素窗口中的所有第二像素点及其行相邻的像素点组成一个像素对；
第二像素点为中心窗口内的另一部分像素点。示例性的，第二像素点为所述四个Bayer格式像素点中的红色点R点或者蓝色点B点。
像素窗口中心四个点中的红色点的降噪是基于红色点所在行的所有像素点进行的，共有14(宽度)x5(高度)个点，如图6所示，可以把行相邻的R和Gr看成一对，一共有7x5个这样的“像素对”。
像素窗口中心四个点中的蓝色点的降噪是基于蓝色点所在行的所有像素点进行的，共有14(宽度)x5(高度)个点，如图7所示，可以把行相邻的B和Gb看成一对，一共有7x5个这样的“像素对”。
步骤204：对每个像素对求和，利用和值构建第二像素矩阵；对每个像素对求差，利用差值构建第三像素矩阵；
步骤205：对所述第二像素矩阵和第三像素矩阵进行所述降噪运算，得到降噪后的和值和差值；
这里，对第二像素矩阵和第三像素矩阵分别执行降噪处理。在一些实施例中，所述降噪运算包括：对像素矩阵中每个像素点先取绝对值，再取开平方，得到像素矩阵中心点的开平方结果sqrt_abs_cen和像素矩阵非中心点的开平方结果sqrt_abs_ref；将每个非中心点sqrt_abs_ref减去中心点sqrt_abs_cen并取绝对值，得到差值ABSDIFF；确定ABSDIFF小于第一阈值的目标非中心点以及目标非中心点个数；将全部目标非中心点进行累加后除以所述目标非中心点个数，得到降噪运算结果。
示例性的，对图6中R点的降噪步骤如下：
步骤1：由于R和Gr总是成对出现，对35对R和Gr分别求和，得到35个(R+Gr)值，此时把(R+Gr)看做一个像素，那么一共有35个这样的像素，这跟Gr和Gb降噪的像素矩阵完全一样，采用相同的降噪运算，可以得到降噪后的(R+Gr)；
步骤2：由于R和Gr总是成对出现，对35对R和Gr分别求差，得到35个(R-Gr)值，此时把(R-Gr)看做一个像素，那么一共有35个这样的像素，这跟Gr和Gb降噪的像素矩阵完全一样，采用相同的降噪运算，可以得到降噪后的(R-Gr)；
步骤3：把步骤1中求得的降噪后(R+Gr)和步骤2中求得的降噪后(R-Gr)这两个数值再次求平均，以此时的计算结果替换中心的R点，完成降噪。
对图7中B点的降噪步骤如下：
步骤1：由于B和Gb总是成对出现，对35对B和Gb分别求和，得到35个(B+Gb)值，此时把(B+Gb)看做一个像素，那么一共有35个这样的像素，这跟Gr和Gb降噪的颜色矩阵完全一样，采用相同的算法，可以得到降噪后的(B+Gb)；
步骤2：由于B和Gb总是成对出现，对35对B和Gb分别求差，得到35个(B-Gb)值，此时把(B-Gb)看做一个像素，那么一共有35个这样的像素，这跟Gr和Gb降噪的颜色矩阵完全一样，采用相同的算法，可以得到降噪后的(B-Gb)；
步骤3：把步骤1中求得的降噪后的(B+Gb)和步骤2中求得的降噪后的(B-Gb)再次求平均，以此时的计算结果替换中心的B点，完成降噪。
需要说明的是，四个点的降噪处理可以同时进行，也可以按照一定顺序先后进行。
步骤206：对降噪后的和值和差值求平均，得到所述中心窗口的降噪后的第二像素点。
在一些实施例中，该方法还包括：所述中心窗口中的四个Bayer格式像素点降噪完成，将所述像素窗口沿水平方向移动两个像素，得到新的像素窗口；对所述新的像素窗口进行降噪处理，直到曝光图像的全部像素点降噪完成。从曝光图像最左侧开始，反复 进行以上操作，直到横向最右侧四个点的降噪完成，14x10的像素窗口向下平移两个像素并回到图像最左侧，开始新的一行的降噪。反复进行以上降噪操作，可以完成整帧图像所有像素点的降噪。
考虑到短曝光图像对HDR图像贡献相比于中曝光图像较小，故对短曝光图像可以采用较简单的降噪处理策略，即通过缩小像素窗口尺寸，减少计算量，简化降噪电路，从而降低降噪成本。示例性的，第一曝光图像为短曝光图像时，像素窗口尺寸可以为6(宽度)x10(高度)。第二曝光图像为中曝光图像时，像素窗口尺寸可以为14(宽度)x10(高度)。
示例性的，如图8所示，对HDR摄像的短曝光图像的降噪运算，以像素窗口是6(宽度)x10(高度)的Bayer格式的窗口，中心窗口的四个Bayer格式像素点RGrGbB是当前像素窗口需要做降噪的点，其他非中心点作可以为降噪参考点参与降噪运算。
短曝光图像像素窗口中心四个点中的绿色点Gr的降噪是以Gr为中心的3(宽度)x5(高度)个Gr点组成的像素矩阵进行的，如图9所示。对Gr点3x5像素矩阵的降噪运算的步骤可以参考中曝光图像中Gr点的降噪运算。
绿色点Gb的降噪是以Gb为中心的3(宽度)x5(高度)个Gr点组成的像素矩阵进行的，如图10所示。对Gb点3x5像素矩阵的降噪运算的步骤可以参考中曝光图像中Gb点的降噪运算。
短曝光图像像素窗口中心四个点中的红色点的降噪是基于红色点所在行的所有像素点进行的，共有6(宽度)x5(高度)个点，如图11所示，可以把行相邻的R和Gr看成一对，一共有3x5个这样的“像素对”。3x5像素矩阵的降噪运算的步骤可以参考中曝光图像中R点的降噪运算。
短曝光图像像素窗口中心四个点中的蓝色点的降噪是基于蓝色点所在行的所有像素点进行的，共有6(宽度)x5(高度)个点，如图12所示，可以把行相邻的B和Gb看成一对，一共有3x5个这样的“像素对”。3x5像素矩阵的降噪运算的步骤可以参考中曝光图像中B点的降噪运算。
同样，完成了短曝光像素窗口中心四个点的降噪。接下来，把6x10的像素窗口在短曝光图像中水平向右平移两个像素，构建了新的6x10像素窗口，采用以上所述降噪的方法，开始对新的6x10像素窗口中心四个点进行降噪。反复进行以上操作，直到横向最右侧四个点的降噪完成，6x10的像素窗口向下平移两个像素并回到图像最左侧，开始新的一行的降噪。反复进行以上降噪操作，可以完成整帧图像所有像素点的降噪。
需要说明的是，若需要对长曝光图像进行降噪处理，可以采用与短曝光图像和中曝光图像的相同的降噪运算方法，只需要依据长曝光图像的降噪需求配置对应的像素窗口尺寸即可。
采用上述技术方案，对HDR摄影的多曝光图像进行处理时，根据曝光时间分别确定用于对至少两幅曝光图像执行降噪处理的像素窗口的尺寸，根据确定的像素窗口的尺寸对至少两幅曝光图像分别执行降噪处理，从而满足不同曝光图像的降噪处理需求有效抑制图像噪声，在保证降噪效果的基础上，节约处理资源，降低降噪成本，优化HDR的降噪方案。
为实现本申请实施例的方法，基于同一发明构思本申请实施例还提供了一种多曝光图像处理装置，如图13所示，该多曝光图像处理装置包括：
获取模块131，配置为获取目标场景在不同曝光时间下拍摄到的至少两幅曝光图像；
降噪模块132，配置为基于所述至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸；基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理；
融合模块133，配置为对经过降噪处理后的所述至少两幅曝光图像进行融合处理，以获得所述目标场景的高动态范围图像。
示例性的，在一些实施例中，所述降噪模块132包括：
像素窗口构建电路141，配置为根据像素窗口的尺寸加载曝光图像的像素窗口；
奇数行提取电路142，配置为从所述像素窗口构建电路141加载的像素窗口中提取奇数行像素；
偶数行提取电路143，配置为从所述像素窗口构建电路141加载的像素窗口中提取偶数行像素；
像素矩阵构建电路144，配置为根据所述奇数行像素或所述偶数行像素构建像素矩阵；
降噪运算电路145，配置为对所述像素矩阵进行降噪运算输出降噪后的像素点。
像素矩阵构建子模块，配置为对于所述像素窗口的中心窗口的四个Bayer格式像素点中的第一像素点，利用所述像素窗口中的全部第一像素点构建第一像素矩阵；
降噪运算子模块，配置为对所述第一像素矩阵进行所述降噪运算，得到所述中心窗口的降噪后的第一像素点；
像素矩阵构建子模块，还用于对于所述像素窗口的中心窗口的四个Bayer格式像素点中的第二像素点，利用所述像素窗口中的所有第二像素点及其行相邻的像素点组成一个像素对；对每个像素对求和，利用和值构建第二像素矩阵；对每个像素对求差，利用差值构建第三像素矩阵；
降噪运算子模块，配置为对所述第二像素矩阵和第三像素矩阵进行所述降噪运算，得到降噪后的和值和差值；对降噪后的和值和差值求平均，得到所述中心窗口的降噪后的第二像素点。
在一些实施例中，所述第一像素点为所述四个Bayer格式像素点中的绿色点Gr点或者绿色点Gb点，所述第二像素点为所述四个Bayer格式像素点中的红色点R点或者蓝色点B点。
在一些实施例中，降噪运算子模块，具体用于对像素矩阵中每个像素点先取绝对值，再取开平方，得到像素矩阵中心点的开平方结果sqrt_abs_cen和像素矩阵非中心点的开平方结果sqrt_abs_ref；将每个非中心点sqrt_abs_ref减去中心点sqrt_abs_cen并取绝对值，得到差值ABSDIFF；确定ABSDIFF小于第一阈值的目标非中心点以及目标非中心点个数；将全部目标非中心点进行累加后除以所述目标非中心点个数，得到降噪运算结果。
示例性的，在一些实施例中，降噪运算子模块，还用于获取配置的权重和偏移量；将所述中心点的开平方结果sqrt_abs_cen和所述权重相乘，得到乘积；将所述乘积和所述偏移量相加，得到所述第一阈值。
示例性的，在一些实施例中，降噪模块132，还用于所述中心窗口中的四个Bayer格式像素点降噪完成，将所述像素窗口沿水平方向移动两个像素，得到新的像素窗口；对所述新的像素窗口进行降噪处理，直到曝光图像的全部像素点降噪完成。
示例性的，在一些实施例中，所述像素窗口的尺寸包括宽度和高度，降噪模块132，还用于使所述像素窗口的宽度和高度中的至少一项与所述曝光时间成正比。
示例性的，在一些实施例中，所述至少两幅曝光图像包括在第一曝光时间下拍摄到的第一曝光图像、在第二曝光时间下拍摄到的第二曝光图像，获取模块131，配置为获取所述目标场景在第三曝光时间下拍摄到的第三曝光图像，其中所述第三曝光时间大于所述第一曝光时间，且大于所述第二曝光时间；
融合模块133，配置为将经过降噪处理后的所述第一曝光图像，经过降噪处理后的所述第二曝光图像，以及未经过降噪处理的所述第三曝光图像进行融合处理，以得到所述目标场景的高动态范围图像。
基于上述多曝光图像处理方法中的曝光图像的降噪方法，本申请实施例还提供了一种降噪电路，示例性的，如图14所示，降噪电路包括：
像素窗口构建电路141，配置为根据曝光图像执行降噪处理的像素窗口的尺寸加载所述曝光图像的像素窗口；
奇数行提取电路142，配置为从所述像素窗口构建电路141加载的像素窗口中提取奇数行像素；
偶数行提取电路143，配置为从所述像素窗口构建电路141加载的像素窗口中提取偶数行像素；
像素矩阵构建电路144，配置为根据所述奇数行像素或所述偶数行像素构建像素矩阵；
降噪运算电路145，配置为对所述像素矩阵进行降噪运算输出降噪后的像素点。
在一些实施例中，降噪电路还包括行缓冲器146，配置为缓存H-1行像素窗口的像素，H为像素窗口的高度，当获取到第H行像素时像素窗口构建电路141加载完整的像素窗口。
在一些实施例中，像素矩阵构建电路144包括：Gr点像素矩阵构建电路、R点像素矩阵构建电路、B点像素矩阵构建电路和Gb点像素矩阵构建电路。
如图3所示，R点和Gr点位于奇数行，Gb点和B点位于偶数行，Gr点像素矩阵构建电路和R点像素矩阵构建电路根据奇数行提取电路142输出的奇数行像素，构建Gr点像素矩阵(即第一像素矩阵)和R点像素矩阵。
B点像素矩阵构建电路和Gb点像素矩阵构建电路根据偶数行提取电路143输出的偶数行像素，构建Gb点像素矩阵(即第一像素矩阵)和B点像素矩阵。
相应的，降噪运算电路145，Gr点降噪运算电路、R点降噪运算电路、B点降噪运算电路和Gb点降噪运算电路，分别对四个点进行降噪运算。
在一些实施例中，降噪电路还包括输出电路147，配置为整理降噪后四个点的输出顺序并输出。
对于像素窗口尺寸为14(宽度)x10(高度)时，降噪电路包括：9行缓冲器和14x10像素窗口构建电路，9行缓冲器用于缓冲前9行像素，当接收到第10行像素，像素窗口构建电路构建14x10像素窗口。利用奇数行提取电路提取14x5像素矩阵，Gr点像素矩阵构建电路提取7x5像素矩阵，R点像素矩阵构建电路直接获取14x5像素矩阵；利用偶数行提取电路提取14x5像素矩阵，B点像素矩阵构建电路提取14x5像素矩阵，Gb点像素矩阵构建电路直接获取3x5像素矩阵。
对于像素窗口尺寸为6(宽度)x10(高度)时，降噪电路包括：9行缓冲器和6x10像素窗口构建电路。利用奇数行提取电路提取6x5像素矩阵，Gr点像素矩阵构建电路提取3x5像素矩阵，R点像素矩阵构建电路直接获取6x5像素矩阵；利用偶数行提取电路提取6x5像素矩阵，B点像素矩阵构建电路提取6x5像素矩阵，Gb点像素矩阵构建电路直接获取3x5像素矩阵。
示例性的，如图15所示，所述降噪运算电路145包括：
取绝对值子电路151，配置为对所述像素矩阵中每个点取绝对值；
开平方子电路152，配置为对所述像素矩阵中每个点的绝对值开平方，得到所述像素矩阵中心点的开平方结果sqrt_abs_cen和像素矩阵非中心点的开平方结果sqrt_abs_ref；
作差取绝对值子电路153，配置为将每个非中心点sqrt_abs_ref减去中心点sqrt_abs_cen并取绝对值，得到差值ABSDIFF；
比较子电路154，配置为将所述ABSDIFF和预设阈值比较，确定ABSDIFF小于第一阈值的目标非中心点以及目标非中心点个数；
第一求平均子电路155，配置为将全部目标非中心点进行累加后除以所述目标非中心点个数，得到降噪后的像素点。
在一些实施例中，所述降噪运算电路145还包括：阈值配置子电路156，配置为获取寄存器配置的权重和偏移量；将所述中心点的开平方结果sqrt_abs_cen和所述权重相乘，得到乘积；将所述乘积和所述偏移量相加，得到所述第一阈值。示例性的，第一阈值等于sqrt_abs_cenGr*reg_weight+reg_offset，其中reg_weight和reg_offset是寄存器配置进来的权重和偏移量。
示例性的，如图16所示，对R点像素矩阵和B点像素矩阵进行降噪运算时，所述降噪运算电路145还包括：像素对求和子电路157和像素对求差子电路158，
所述像素对求和子电路157，配置为对所述像素矩阵构建电路输出的像素矩阵中行相邻的像素对求和，得到第二像素矩阵，并输入到取绝对值子电路；
所述像素对求差子电路158，配置为对所述像素矩阵构建电路输出的像素矩阵中行相邻的像素对求差，得到第三像素矩阵，并输入到所述取绝对值子电路。
由于R和Gr总是成对出现，对35对R和Gr分别求和，得到35个(R+Gr)值，此时把(R+Gr)看做一个像素，那么一共有35个这样的像素，这跟Gr和Gb降噪的像素矩阵完全一样，采用相同的降噪运算，可以得到降噪后的(R+Gr)。
由于R和Gr总是成对出现，对35对R和Gr分别求差，得到35个(R-Gr)值，此时把(R-Gr)看做一个像素，那么一共有35个这样的像素，这跟Gr和Gb降噪的像素矩阵完全一样，采用相同的降噪运算，可以得到降噪后的(R-Gr)。
中间与图15相同，只是在最后增加第二求平均子电路159，配置为对所述第一求平均子电路输出的降噪后的和值和差值求平均，得到所述中心窗口的降噪后的第二像素点。
示例性的，Gr降噪运算电路和Gb降噪运算电路可以为图15所示降噪运算电路，R降噪运算电路和B降噪运算电路可以为图16所示降噪运算电路。
实际应用中，多曝光图像处理装置中的降噪模块可以由上述降噪电路来实现。
本申请还提供了一种多曝光图像处理芯片，如图17所示，所述芯片170包括上述实施例中任一项降噪电路1701。
可选地，该多曝光图像处理芯片170还可以包括输入接口1702。通过输入接口1702与其他设备或芯片进行通信，具体地，可以获取其他设备或芯片发送的信息或数据。
可选地，该多曝光图像处理芯片170还可以包括输出接口1703。通过输出接口1703与其他设备或芯片进行通信，具体地，可以向其他设备或芯片输出信息或数据。
可选地，该多曝光图像处理芯片170可应用于本申请实施例中的电子设备。
应理解，本申请实施例提到的芯片还可以称为系统级芯片，系统芯片，芯片系统或片上系统芯片等。
在实际应用中，上述处理器可以为特定用途集成电路(ASIC，Application Specific Integrated Circuit)、数字信号处理装置(DSPD，Digital Signal Processing Device)、可编程逻辑装置(PLD，Programmable Logic Device)、现场可编程门阵列(Field－Programmable Gate Array，FPGA)、控制器、微控制器、微处理器中的至少一种。可以理解地，对于不同的设备，配置为实现上述处理器功能的电子器件还可以为其它，本申请实施例不作具体限定。
上述存储器可以是易失性存储器(volatile memory)，例如随机存取存储器(RAM，Random-Access Memory)；或者非易失性存储器(non-volatile memory)，例如只读存储器(ROM，Read-Only Memory)，快闪存储器(flash memory)，硬盘(HDD，Hard Disk Drive)或固态硬盘(SSD，Solid-State Drive)；或者上述种类的存储器的组合，并向处理器提供指令和数据。
本申请实施例还提供了电子设备，本申请中描述的电子设备具备拍摄功能，能够拍摄不同曝光时间的曝光图像，电子设备可以包括诸如手机、平板电脑、笔记本电脑、掌上电脑、个人数字助理(Personal Digital Assistant，PDA)、便捷式媒体播放器(Portable Media Player，PMP)、可穿戴设备、相机、智能汽车等。
如图18所示，该电子设备180包括：图像采集装置1801以及前述多曝光图像处理芯片1802。
示例性的，图像采集装置1801，配置为采集目标场景在至少两种曝光时间下拍摄到的至少两张曝光图像；
多曝光图像处理芯片1802，为上述实施例中任一项多曝光图像处理芯片，配置为对不同曝光图像采用不同的降噪处理策略进行降噪处理。
可选地，所述多曝光图像处理芯片1802还用于将降噪后的曝光图像进行融合。
当然，实际应用时，如18所示，该电子设备180中的各个组件通过总线系统1803耦合在一起。可理解，总线系统1803用于实现这些组件之间的连接通信。总线系统1803除包括数据总线之外，还包括电源总线、控制总线和状态信号总线。但是为了清楚说明起见，在图18中将各种总线都标为总线系统1803。
在示例性实施例中，本申请实施例还提供了一种计算机可读存储介质，例如包括计算机程序的存储器，计算机程序可由处理器执行，以完成前述方法的步骤。
本申请实施例还提供了一种计算机程序产品，包括多个指令，所述指令被计算设备执行时实现本申请实施例中多曝光图像处理方法的步骤。
可选的，该计算机程序产品可应用于本申请实施例中的处理器，并且该计算机程序指令使得计算机执行本申请实施例的各个方法中由处理器实现的相应流程，为了简洁，在此不再赘述。
本申请实施例还提供了一种计算机程序。
可选的，该计算机程序可应用于本申请实施例中的处理器，当该计算机程序在计算机上运行时，使得计算机执行本申请实施例的各个方法中由处理器实现的相应流程，为了简洁，在此不再赘述。
应当理解，在本申请使用的术语是仅仅出于描述特定实施例的目的，而非旨在限制本申请。在本申请和所附权利要求书中所使用的单数形式的“一种”、“所述”和“该”也旨在包括多数形式，除非上下文清楚地表示其他含义。还应当理解，本文中使用的术语“和/或”是指并包含一个或多个相关联的列出项目的任何或所有可能组合。本申请中表述“具有”、“可以具有”、“包括”和“包含”、或者“可以包括”和“可以包含”在本文中可以用于指示存在对应的特征(例如，诸如数值、功能、操作或组件等元素)，但不排除附加特征的存在。
应当理解，尽管在本申请可能采用术语第一、第二、第三等来描述各种信息，但这些信息不应限于这些术语。这些术语仅用来将同一类型的信息彼此区分开，不必用于描述特定的顺序或先后次序。例如，在不脱离本发明范围的情况下，第一信息也可以被称为第二信息，类似地，第二信息也可以被称为第一信息。
本申请实施例所记载的技术方案之间，在不冲突的情况下，可以任意组合。
在本申请所提供的几个实施例中，应该理解到，所揭露的方法、装置和设备，可以 通过其它的方式实现。以上所描述的实施例仅仅是示意性的，例如，单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，如：多个单元或组件可以结合，或可以集成到另一个系统，或一些特征可以忽略，或不执行。另外，所显示或讨论的各组成部分相互之间的耦合、或直接耦合、或通信连接可以是通过一些接口，设备或单元的间接耦合或通信连接，可以是电性的、机械的或其它形式的。
上述作为分离部件说明的单元可以是、或也可以不是物理上分开的，作为单元显示的部件可以是、或也可以不是物理单元，即可以位于一个地方，也可以分布到多个网络单元上；可以根据实际的需要选择其中的部分或全部单元来实现本实施例方案的目的。
另外，在本申请各实施例中的各功能单元可以全部集成在一个处理单元中，也可以是各单元分别单独作为一个单元，也可以两个或两个以上单元集成在一个单元中；上述集成的单元既可以采用硬件的形式实现，也可以采用硬件加软件功能单元的形式实现。
以上所述，仅为本申请的具体实施方式，但本申请的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本申请揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本申请的保护范围之内。
工业实用性
本申请实施例提供了一种多曝光图像处理方法、装置及降噪电路，该方法对HDR摄影的多曝光图像进行处理时，根据曝光时间分别确定用于对至少两幅曝光图像执行降噪处理的像素窗口的尺寸，根据确定的像素窗口的尺寸对至少两幅曝光图像分别执行降噪处理，从而满足不同曝光图像的降噪处理需求有效抑制图像噪声，在保证降噪效果的基础上，节约处理资源，降低降噪成本，优化HDR的降噪方案。

Claims (20)

  1. 一种多曝光图像处理方法，其中，所述方法包括：
    获取目标场景在不同曝光时间下拍摄到的至少两幅曝光图像；
    基于所述至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸；
    基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理；
    对经过降噪处理后的所述至少两幅曝光图像进行融合处理，以获得所述目标场景的高动态范围图像。
  2. 如权利要求1所述的多曝光图像处理方法，其中，所述至少两幅曝光图像为Bayer格式图像，所述基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理包括：
    对于所述像素窗口的中心窗口的四个Bayer格式像素点中的第一像素点，利用所述像素窗口中的全部第一像素点构建第一像素矩阵；
    对所述第一像素矩阵进行降噪运算，得到所述中心窗口的降噪后的第一像素点；
    对于所述像素窗口的中心窗口的四个Bayer格式像素点中的第二像素点，利用所述像素窗口中的所有第二像素点及其行相邻的像素点组成一个像素对；
    对每个像素对求和，利用和值构建第二像素矩阵；
    对每个像素对求差，利用差值构建第三像素矩阵；
    对所述第二像素矩阵和所述第三像素矩阵进行所述降噪运算，得到降噪后的和值和差值；
    对降噪后的和值和差值求平均，得到所述中心窗口的降噪后的第二像素点。
  3. 如权利要求2所述的多曝光图像处理方法，其中，
    所述第一像素点为所述四个Bayer格式像素点中的绿色点Gr点或者绿色点Gb点，
    所述第二像素点为所述四个Bayer格式像素点中的红色点R点或者蓝色点B点。
  4. 如权利要求2或3所述的多曝光图像处理方法，其中，所述降噪运算包括：
    对像素矩阵中每个像素点先取绝对值，再取开平方，得到像素矩阵中心点的开平方结果sqrt_abs_cen和像素矩阵非中心点的开平方结果sqrt_abs_ref；
    将每个非中心点sqrt_abs_ref减去中心点sqrt_abs_cen并取绝对值，得到差值ABSDIFF；
    确定ABSDIFF小于第一阈值的目标非中心点以及目标非中心点个数；
    将全部目标非中心点进行累加后除以所述目标非中心点个数，得到降噪运算结果。
  5. 如权利要求4所述的多曝光图像处理方法，其中，所述方法还包括：
    获取配置的权重和偏移量；
    将所述中心点的开平方结果sqrt_abs_cen和所述权重相乘，得到乘积；
    将所述乘积和所述偏移量相加，得到所述第一阈值。
  6. 如权利要求2或3所述的多曝光图像处理方法，其中，所述方法还包括：
    所述中心窗口中的四个Bayer格式像素点降噪完成，将所述像素窗口沿水平方向移动两个像素，得到新的像素窗口；
    对所述新的像素窗口进行降噪处理，直到曝光图像的全部像素点降噪完成。
  7. 如权利要求1-3中的任一项所述的多曝光图像处理方法，其中，所述像素窗口的尺寸包括宽度和高度，所述基于所述至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸包括：
    使所述像素窗口的宽度和高度中的至少一项与所述曝光时间成正比。
  8. 如权利要求1-3中的任一项所述的多曝光图像处理方法，其中，所述至少两幅曝光图像包括在第一曝光时间下拍摄到的第一曝光图像、在第二曝光时间下拍摄到的第二曝光图像，所述方法还包括：
    获取所述目标场景在第三曝光时间下拍摄到的第三曝光图像，其中所述第三曝光时间大于所述第一曝光时间，且大于所述第二曝光时间；
    将经过降噪处理后的所述第一曝光图像，经过降噪处理后的所述第二曝光图像，以及未经过降噪处理的所述第三曝光图像进行融合处理，以得到所述目标场景的高动态范围图像。
  9. 一种多曝光图像处理装置，其中，包括：
    获取模块，配置为获取目标场景在不同曝光时间下拍摄到的至少两幅曝光图像；
    降噪模块，配置为基于所述至少两幅曝光图像的曝光时间，分别确定用于对所述至少两幅曝光图像执行降噪处理的像素窗口的尺寸；基于所确定的像素窗口的尺寸对所述至少两幅曝光图像分别执行降噪处理；
    融合模块，配置为对经过降噪处理后的所述至少两幅曝光图像进行融合处理，以获得所述目标场景的高动态范围图像。
  10. 一种降噪电路，其中，所述降噪电路包括：
    像素窗口构建电路，配置为根据曝光图像执行降噪处理的像素窗口的尺寸加载所述曝光图像的像素窗口；
    奇数行提取电路，配置为从所述像素窗口构建电路加载的像素窗口中提取奇数行像素；
    偶数行提取电路，配置为从所述像素窗口构建电路加载的像素窗口中提取偶数行像素；
    像素矩阵构建电路，配置为根据所述奇数行像素或所述偶数行像素构建像素矩阵；
    降噪运算电路，配置为对所述像素矩阵进行降噪运算输出降噪后的像素点。
  11. 如权利要求10所述的降噪电路，其中，所述降噪运算电路包括：
    取绝对值子电路，配置为对所述像素矩阵中每个点取绝对值；
    开平方子电路，配置为对所述像素矩阵中每个点的绝对值开平方，得到所述像素矩阵中心点的开平方结果sqrt_abs_cen和像素矩阵非中心点的开平方结果sqrt_abs_ref；
    作差取绝对值子电路，配置为将每个非中心点sqrt_abs_ref减去中心点sqrt_abs_cen并取绝对值，得到差值ABSDIFF；
    比较子电路，配置为将所述ABSDIFF和预设阈值比较，确定ABSDIFF小于第一阈值的目标非中心点以及目标非中心点个数；
    第一求平均子电路，配置为将全部目标非中心点进行累加后除以所述目标非中心点个数，得到降噪后的像素点。
  12. 如权利要求11所述的降噪电路，其中，所述降噪运算电路还包括：
    阈值配置子电路，配置为获取寄存器配置的权重和偏移量；将所述中心点的开平方结果sqrt_abs_cen和所述权重相乘，得到乘积；将所述乘积和所述偏移量相加，得到所述第一阈值。
  13. 如权利要求11或12所述的降噪电路，其中，所述像素矩阵构建电路包括：Gr点像素矩阵构建电路和Gb点像素矩阵构建电路；
    所述Gr点像素矩阵构建电路，配置为根据所述奇数行像素中的Gr点像素构建Gr点像素矩阵；
    所述Gb点像素矩阵构建电路，配置为根据所述偶数行像素中的Gb点像素构建Gb点像素矩阵。
  14. 如权利要求13所述的降噪电路，其中，所述像素矩阵构建电路包括：R点像素矩阵构建电路和B点像素矩阵构建电路；
    所述R点像素矩阵构建电路，配置为根据所述奇数行像素中的R点像素构建R点像素矩阵；
    所述B点像素矩阵构建电路，配置为根据所述偶数行像素中的B点像素构建B点像素矩阵。
  15. 根据权利要求14所述的降噪电路，其中，所述降噪运算电路还包括：像素对求和子电路和像素对求差子电路，
    所述像素对求和子电路，配置为对所述像素矩阵构建电路输出的像素矩阵中行相邻的像素对求和，得到第二像素矩阵，并输入到所述取绝对值子电路；
    所述像素对求差子电路，配置为对所述像素矩阵构建电路输出的像素矩阵中行相邻的像素对求差，得到第三像素矩阵，并输入到所述取绝对值子电路。
  16. 如权利要求15所述的降噪电路，其中，所述降噪运算电路还包括：第二求平均子电路，配置为对所述第一求平均子电路输出的降噪后的和值和差值求平均，得到所述中心窗口的降噪后的第二像素点。
  17. 一种多曝光图像处理芯片，其中，所述芯片包括如权利要求10-16任一项所述的降噪电路。
  18. 一种电子设备，其中，所述电子设备包括：图像采集装置以及如权利要求17所述的多曝光图像处理芯片。
  19. 一种计算机可读存储介质，其上存储有计算机程序，其中，所述计算机程序被处理器执行时实现权利要求1至8任一项所述方法的步骤。
  20. 一种计算机程序产品，包括多个指令，所述指令被计算设备执行时实现如权利要求1至8任一项所述的方法。
PCT/CN2022/119813 2021-09-26 2022-09-20 多曝光图像处理方法、装置及降噪电路 WO2023045907A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111129165.5 2021-09-26
CN202111129165.5A CN115942125A (zh) 2021-09-26 2021-09-26 一种多曝光图像处理方法、装置及降噪电路

Publications (1)

Publication Number Publication Date
WO2023045907A1 true WO2023045907A1 (zh) 2023-03-30

Family

ID=85720063

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/119813 WO2023045907A1 (zh) 2021-09-26 2022-09-20 多曝光图像处理方法、装置及降噪电路

Country Status (2)

Country Link
CN (1) CN115942125A (zh)
WO (1) WO2023045907A1 (zh)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103116873A (zh) * 2013-03-20 2013-05-22 中国科学院深圳先进技术研究院 图像降噪方法
CN103327220A (zh) * 2013-05-29 2013-09-25 山西绿色光电产业科学技术研究院（有限公司） 低照度Bayer图像上以绿色通道为导向的去噪方法
CN107635098A (zh) * 2017-10-30 2018-01-26 广东欧珀移动通信有限公司 高动态范围图像噪声去除方法、装置及设备
US20190213722A1 (en) * 2016-02-05 2019-07-11 Sony Corporation Image processing apparatus and image processing method
CN110062160A (zh) * 2019-04-09 2019-07-26 Oppo广东移动通信有限公司 图像处理方法和装置
CN110290289A (zh) * 2019-06-13 2019-09-27 Oppo广东移动通信有限公司 图像降噪方法、装置、电子设备以及存储介质

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103116873A (zh) * 2013-03-20 2013-05-22 中国科学院深圳先进技术研究院 图像降噪方法
CN103327220A (zh) * 2013-05-29 2013-09-25 山西绿色光电产业科学技术研究院（有限公司） 低照度Bayer图像上以绿色通道为导向的去噪方法
US20190213722A1 (en) * 2016-02-05 2019-07-11 Sony Corporation Image processing apparatus and image processing method
CN107635098A (zh) * 2017-10-30 2018-01-26 广东欧珀移动通信有限公司 高动态范围图像噪声去除方法、装置及设备
CN110062160A (zh) * 2019-04-09 2019-07-26 Oppo广东移动通信有限公司 图像处理方法和装置
CN110290289A (zh) * 2019-06-13 2019-09-27 Oppo广东移动通信有限公司 图像降噪方法、装置、电子设备以及存储介质

Also Published As

Publication number Publication date
CN115942125A (zh) 2023-04-07

Similar Documents

Publication Publication Date Title
US11301728B2 (en) Image processing using a neural network system
US11151704B2 (en) Apparatus and methods for artifact detection and removal using frame interpolation techniques
US9007488B2 (en) Systems and methods for generating interpolated high-dynamic-range images
US8824831B2 (en) Advanced noise reduction in digital cameras
CN109417613B (zh) 具有多个连续红外滤波单元的图像传感器方法和装置
US20190355142A1 (en) Image processing
WO2013006512A2 (en) Image signal processor architecture optimized for low-power, processing flexibility, and user experience
CN107431743B (zh) 数字原始图像帧的按比例缩小
JP4397079B2 (ja) イメージセンサにおける緑色画素の平均値を利用する自動露出調節装置及びその方法
CN114331916B (zh) 图像处理方法及电子设备
CN112703728B (zh) 用于处理数字图像的方法、装置和系统
WO2023045907A1 (zh) 多曝光图像处理方法、装置及降噪电路
CN103458249A (zh) 图像处理方法及装置
CN113298740A (zh) 一种图像增强方法、装置、终端设备及存储介质
US20220366588A1 (en) Electronic device for generating depth information of region of interest and operation method thereof
Chao et al. Pyramid architecture for 3840 x 2160 quad full high definition 30 frames/s video acquisition
Pan et al. FPGA Acceleration of Color Interpolation Algorithm Based on Gradient and Color Difference
US9659346B2 (en) Image processing apparatus, image processing method, solid-state imaging device, and electronic apparatus configured to calculate a pixel value of a target position in accordance with a weighted value of each pixel on a candidate line of a plurality of candidate lines
WO2017084011A1 (zh) 一种视频平滑方法及装置
JP2017041732A (ja) 画像処理装置、画像処理方法およびプログラム
Wu et al. Joint Demosaicing and Denoising with Gradient Guidance in Quad Bayer CFA
KR102071460B1 (ko) 화상 처리 장치 및 화상 처리 방법
Jin et al. A 12-bit 4928× 3264 pixel CMOS image signal processor for digital still cameras
Khalifat et al. An efficient implementation of the Adams-Hamilton’s demosaicing algorithm in FPGAs
US8437546B2 (en) Image processing apparatus and method of controlling the same