Image sensor signal-to-noise ratio evaluation method based on specific flat area

Abstract

The present invention relates to the field of image sensor technology, and specifically to a method for evaluating the signal-to-noise ratio of an image sensor based on a specific flat area. The method comprises the following steps: S1, constructing a shooting lighting environment and scene containing a flat area; S2, acquiring a Bayer array original image using a camera; S3, reading the Bayer array original image using a computer device and selecting a coordinate range of a flat area in the Bayer array original image; S4, separating the grayscale values ​​of each color channel according to the flat area coordinates and the Bayer array arrangement; S5, calculating the mean and standard deviation of the grayscale values ​​of each color channel to obtain the signal-to-noise ratio of the selected flat area; and optionally, S6, comparing the signal-to-noise ratios of the same average area of ​​different image sensors under the same conditions to estimate their performance differences. The method of the present invention does not require excessive sophisticated equipment or complicated steps, and can quickly determine the basic noise performance of an image sensor, particularly for comparative evaluation between two or more image sensors.

Description

Technical Field

[0001] The present invention relates to the technical field of image sensors, and in particular to a method for evaluating the signal-to-noise ratio of an image sensor based on a specific flat area. Background Art

[0002] With the advancement of science and technology, the application of various camera devices is becoming increasingly widespread. To meet the needs of everyday photography, driving recorders, security monitoring, and other diverse applications, image quality requirements are becoming increasingly stringent. This requires that camera equipment, during development and testing, ensure stable, high-quality images. Image noise is a key indicator of image quality, and a simple and convenient method to measure this metric is urgently needed by relevant practitioners.

[0003] Many factors determine the final image quality of camera equipment. From the source, the quality of the original image output by the image sensor plays a crucial role in the final image. The "Image Sensor and Camera Performance Test Standard" (EMVA1288), developed by the European Machine Vision Association (EMVA), comprehensively evaluates image sensor performance in areas such as noise, serving as a guide within the industry. However, the EMVA1288 standard requires sophisticated equipment, cumbersome procedures, and high costs, making it difficult for small-scale practitioners to implement the evaluation process.

[0004] Therefore, a method is needed to conveniently evaluate image sensors, improve efficiency and save costs. Summary of the Invention

[0005] In order to solve the above problems, the present invention provides an image sensor signal-to-noise ratio evaluation method based on a specific flat area, which can quickly determine the basic performance of the image sensor in terms of noise.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] The method for evaluating the signal-to-noise ratio of an image sensor based on a specific flat area includes the following steps:

[0008] S1, constructing the shooting lighting environment and scene with flat areas;

[0009] Shooting lighting environment and scene can be achieved by the following means

[0010] 1) It has a panel-type light source with uniform light emission, which is generally available to those working with camera equipment;

[0011] 2) A chart or wall with uniform reflective material. These requirements are usually met by camera equipment developers and testers.

[0012] In particular, considering that the noise of the image sensor is more obvious in a low-light environment, in actual operation, it is preferred that the shooting environment is in a relatively dark condition.

[0013] S2, using a camera to obtain a Bayer array original image. When obtaining the Bayer array original image, in addition to recording the grayscale value information of the original image, it is also necessary to record relevant information such as the Bayer array of the original image;

[0014] S3, using a computer device to read the above Bayer array original image, and select the coordinate range of the flat area in the Bayer array original image;

[0015] S4, separating the grayscale value of each color channel according to the flat area coordinates and the Bayer array arrangement of the Bayer array original image;

[0016] S5, calculating the mean and standard deviation of the grayscale values ​​of each color channel, obtaining the signal-to-noise ratio of the selected flat area, and evaluating the image sensor performance based on the signal-to-noise ratio value.

[0017] Image sensors with larger signal-to-noise ratios have better performance.

[0018] Optionally, in one embodiment of the present invention, the Bayer array in step S3 is a 2*2 array, the array includes two green channels Gr and Gb, one red channel R, and one blue channel B, for a total of four color channels. The four color channels have a total of four permutations and combinations, and the Bayer array original image is any one of the four permutations and combinations.

[0019] The four permutation and combination methods are as follows:

[0020]

[0021] Optionally, in an embodiment of the present invention, when selecting the flat area coordinate range in step S3, the Bayer array of the selected flat area is the same as the Bayer array of the Bayer array original image.

[0022] Optionally, in one embodiment of the present invention, step S3 of obtaining the flat area coordinates specifically involves recording the number of m*n pixels in the original Bayer array image rows and columns, with the starting point being 1,1, the coordinates of the xth column and yth row being recorded as x,y, and the last point being recorded as m,n. When selecting the flat area coordinates, a rectangular area is obtained, with the starting point coordinates being x1,y1 and the ending point being x2,y2. The starting and ending point coordinates are then substituted into the following formula:

[0023] x1′=floor(x1 / 2)*2+1

[0024] y1′=floor(y1 / 2)*2+1

[0025] x2′=floor(x2 / 2)*2

[0026] y2′=floor(y2 / 2)*2.

[0027] In the above formula, floor represents the integer portion of the result in the parentheses. The selected coordinate range is calculated as described above to obtain the coordinates of the plane area aligned with the Bayer array of the original Bayer array image: the starting point (x1′, y1′) in the upper left and the ending point (x2′, y2′) in the lower right of the plane area.

[0028] Optionally, in an embodiment of the present invention, step S4 specifically comprises: creating four new tables, and sequentially filling in the grayscale values ​​of the four color channels in the flat area until each pixel point is sequentially filled into the designated channel table.

[0029] Optionally, in one embodiment of the present invention, step S5 further includes the following steps:

[0030] S501, calculate the mean of the grayscale values. Specifically, a channel has a grayscale value, the grayscale value is z1, z2, ..., za, then the mean Z 均值 Obtained by the following formula:

[0031]

[0032] Z 均值 It is the value reflecting the signal strength of the channel.

[0033] S502, calculate the variance of the gray value, specifically, the variance is recorded as Z 方差 , then the variance is obtained by the following formula:

[0034]

[0035] S503, calculate the standard deviation of the gray value. Specifically, the standard deviation represents the arithmetic square root of the variance, and the standard deviation is recorded as Z 标准差 , the standard deviation is obtained by the following formula:

[0036]

[0037] Z 标准差 It is a value reflecting the noise intensity of the channel.

[0038] S504: Based on the results of steps S501-S503, the quotient of the mean and the standard deviation is calculated to obtain the signal-to-noise ratio. Specifically, the signal-to-noise ratio of the color channel is denoted as SNR. The signal-to-noise ratio is obtained by the following formula:

[0039]

[0040] Repeating S501 to S504 can obtain the signal-to-noise ratios corresponding to the four color channels. By comparing the differences in the signal-to-noise ratios of the four color channels, the performance of the image sensor in terms of noise at different colors can be intuitively obtained, and the larger the value, the better.

[0041] The method also includes an optional step S6, which specifically comprises: comparing the signal-to-noise ratios of the same average area of ​​different image sensors under the same conditions to estimate their performance differences. This step is mainly applicable when comparing and evaluating two image sensors or different output modes of the same image sensor.

[0042] Repeat steps S1 to S5 to ensure that the signal-to-noise ratios of two or more image sensors are calculated under the same conditions. The same conditions refer to:

[0043] (1) Same lighting conditions and scenes;

[0044] (2) The image sensors operate at the same exposure time and output magnification;

[0045] (3) The image areas captured by the original images are the same or similar;

[0046] (4) Select the same flat area.

[0047] The signal-to-noise ratio is an important indicator for evaluating the performance of image sensors in terms of noise. If the two results are close, it means that the two have similar performance in this regard; if one of them is larger, it means that the image sensor corresponding to the larger one has better performance.

[0048] Beneficial effects of the present invention

[0049] The image sensor signal-to-noise ratio evaluation method based on a specific flat area of ​​the present invention does not require excessive precision equipment and cumbersome steps. It is applicable to the cross-evaluation of multiple output modes of the same image sensor or multiple image sensors, and facilitates practitioners related to the research and development and testing of camera equipment to quickly and cheaply understand the basic performance of the image sensors used. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1This is a flow chart of the method for evaluating the signal-to-noise ratio of an image sensor based on a specific flat area provided in Example 1 of the present invention;

[0052] Figure 2 Flowchart for calculating image signal, noise and signal-to-noise ratio provided by Example 1 of the present invention;

[0053] Figure 3 Graphs of actual test data from the method of the present invention; DETAILED DESCRIPTION

[0054] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and are not to be understood as limiting the present invention. Therefore, the description is for illustrative purposes only. Therefore, the practice of the inventive concept herein is not limited solely to the example embodiments described herein and the illustrations in the accompanying drawings. In addition, the drawings are not limiting.

[0055] Example 1

[0056] The existing evaluation method requires more sophisticated equipment, and the evaluation process is relatively cumbersome. In order to improve the evaluation efficiency and reduce the cost, a relatively convenient evaluation method is designed. The specific plan is as follows: Figure 1 As shown, the image sensor signal-to-noise ratio evaluation method based on a specific flat area includes the following steps:

[0057] S1, constructing the shooting lighting environment and scene with flat areas;

[0058] Shooting lighting environment and scene can be achieved by the following means

[0059] 1) It has a panel-type light source with uniform light emission, which is generally available to those working with camera equipment;

[0060] 2) A chart or wall with uniform reflective material. These requirements are usually met by camera equipment developers and testers.

[0061] In particular, considering that the noise of the image sensor is more obvious in a low-light environment, in actual operation, it is preferred that the shooting environment is in a relatively dark condition.

[0062] S2, using a camera to obtain a Bayer array original image. When obtaining the Bayer array original image, in addition to recording the grayscale value information of the original image, it is also necessary to record relevant information such as the Bayer array of the original image;

[0063] S3, using a computer device to read the above Bayer array original image, and select the coordinate range of the flat area in the Bayer array original image;

[0064] The Bayer array in step S3 is a 2*2 array, which includes two green channels Gr and Gb, one red channel R, and one blue channel B, for a total of four color channels. There are four permutations and combinations of the four color channels. The original Bayer array image is any one of the four permutations and combinations.

[0065] The four permutation and combination methods are as follows:

[0066]

[0067] When selecting the coordinate range of the flat area in step S3, the Bayer array of the selected flat area is the same as the Bayer array of the Bayer array original image.

[0068] Step S3 obtains the flat area coordinates as follows: record the number of m*n pixels in the original Bayer array image, with the starting point being 1,1. The coordinates of the xth column and yth row can be recorded as x,y, and the last point can be recorded as m,n. When selecting the flat area coordinates, a rectangular area is obtained, with the starting point coordinates being x1,y1 and the ending point being x2,y2. Substitute the starting and ending point coordinates into the following formula:

[0069] x1′=floor(x1 / 2)*2+1

[0070] y1′=floor(y1 / 2)*2+1

[0071] x2′=floor(x2 / 2)*2

[0072] y2′=floor(y2 / 2)*2

[0073] In the above formula, floor represents the integer portion of the result in the parentheses. The selected coordinate range is calculated as described above to obtain the coordinates of the plane area aligned with the Bayer array of the original Bayer array image: the starting point (x1′, y1′) in the upper left and the ending point (x2′, y2′) in the lower right of the plane area.

[0074] S4, separating the grayscale value of each color channel according to the flat area coordinates and the Bayer array arrangement of the Bayer array original image;

[0075] Step S4 specifically includes: creating four new tables, and filling in the grayscale values ​​of the four color channels in the flat area in sequence until each pixel point is filled into the designated channel table in order.

[0076] S5, calculating the mean and standard deviation of the grayscale values ​​of each color channel, obtaining the signal-to-noise ratio of the selected flat area, and evaluating the image sensor performance based on the signal-to-noise ratio value.

[0077] Image sensors with larger signal-to-noise ratios have better performance.

[0078] like Figure 2 As shown, step S5 further includes the following steps:

[0079] S501, calculate the mean of the grayscale values. Specifically, a channel has a grayscale value, and the grayscale values ​​are z1, z2, ..., z a , then the mean Z 均值 Obtained by the following formula:

[0080]

[0081] Z 均值 It is the value reflecting the signal strength of the channel.

[0082] S502, calculate the variance of the gray value, specifically, the variance is recorded as Z 方差 , then the variance is obtained by the following formula:

[0083]

[0084] S503, calculate the standard deviation of the gray value. Specifically, the standard deviation represents the arithmetic square root of the variance, and the standard deviation is recorded as Z 标准差 , the standard deviation is obtained by the following formula:

[0085]

[0086] Z 标准差 It is a value reflecting the noise intensity of the channel.

[0087] S504: Based on the results of steps S501-S503, the quotient of the mean and the standard deviation is calculated to obtain the signal-to-noise ratio. Specifically, the signal-to-noise ratio of the color channel is denoted as SNR. The signal-to-noise ratio is obtained by the following formula:

[0088]

[0089] Repeating S501 to S504 can obtain the signal-to-noise ratios corresponding to the four color channels. By comparing the differences in the signal-to-noise ratios of the four color channels, the performance of the image sensor in terms of noise at different colors can be intuitively obtained, and the larger the value, the better.

[0090] The method also includes an optional step S6, which specifically comprises: comparing the signal-to-noise ratios of the same average area of ​​different image sensors under the same conditions to estimate their performance differences. This step is mainly applicable when comparing and evaluating two image sensors or different output modes of the same image sensor.

[0091] Repeat steps S1 to S5 to ensure that the signal-to-noise ratios of two or more image sensors are calculated under the same conditions. The same conditions refer to:

[0092] (1) Same lighting conditions and scenes;

[0093] (2) The image sensors operate at the same exposure time and output magnification;

[0094] (3) The image areas captured by the original images are the same or similar;

[0095] (4) Select the same flat area.

[0096] The signal-to-noise ratio is an important indicator for evaluating the performance of image sensors in terms of noise. If the two results are close, it means that the two have similar performance in this regard; if one of them is larger, it means that the image sensor corresponding to the larger one has better performance.

[0097] The solution also provides a test scenario, as follows:

[0098] In this scene, the gray wall in front of the camera is a commonly used 18-degree neutral gray wall. The wall's characteristic is uniform reflection, so the image sensor obtains an original image with a flat area.

[0099] Behind the camera equipment, a uniformly emitting panel light source is used to ensure uniform reflection of the above-mentioned gray wall;

[0100] Select the plane part on the neutral gray wall in the image as the plane area coordinate range;

[0101] It should be noted that the original image output by the image sensor has not undergone subsequent image signal processing (ISP). The grayscale value of the green channel is greater than that of the red and blue channels. Therefore, the gray wall appears green in the image displayed during the evaluation process. This does not affect the final evaluation results of the image sensor.

[0102] like Figure 3 As shown, in this figure, actual test data of two different image sensors are provided, and the test data includes two different output modes of the same image sensor and data of the same output mode of the same sensor.

[0103] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any form. Although the present invention has been disclosed as above in terms of preferred embodiments, they are not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for evaluating the signal-to-noise ratio of an image sensor based on a specific flat area, characterized by: The following steps are involved: S1, constructing the shooting lighting environment and scene with flat areas; S2, using a camera to obtain the original image of the Bayer array; S3, using a computer device to read the above Bayer array original image, and select the coordinate range of the flat area in the Bayer array original image; S4, separating the grayscale value of each color channel according to the flat area coordinates and the Bayer array arrangement of the Bayer array original image; S5, calculating the mean and standard deviation of the grayscale values ​​of each color channel, obtaining the signal-to-noise ratio of the selected flat area, and evaluating the image sensor performance based on the signal-to-noise ratio value; The Bayer array in step S3 is a 2*2 array, including two green channels, one red channel, and one blue channel, for a total of four color channels. The four color channels have four permutations and combinations, and the original Bayer array image is any one of the four permutations and combinations; When selecting the coordinate range of the flat area in step S3, the Bayer array of the selected flat area is the same as the Bayer array of the Bayer array original image; The step S3 of obtaining the flat area coordinates is as follows: record the number of m*n pixels in the original Bayer array image, with the starting point being 1,1, the coordinates of the xth column and yth row being x,y, and the last point being m,n. When selecting the flat area coordinates, a rectangular area is obtained, with the starting point coordinates being x1,y1 and the ending point being x2,y2. Substitute the starting point coordinates and the ending point coordinates into the following formula: x1′=floor(x1 / 2)*2+1 y1′=floor(y1 / 2)*2+1 x2′=floor(x2 / 2)*2 y2′=floor(y2 / 2)*2; Wherein, the floor represents taking the integer part of the result in the brackets; The selected coordinate range is calculated using the above formula to obtain the coordinates of the plane area aligned with the Bayer array of the Bayer array original image: the starting point (x1′, y1′) located in the upper left and the ending point (x2′, y2′) located in the lower right of the plane area.

2. The method for evaluating the signal-to-noise ratio of an image sensor based on a specific flat area according to claim 1, wherein: The step S4 specifically includes: creating four new tables, and sequentially filling in the grayscale values ​​of the four color channels in the flat area until each pixel point is sequentially filled into the designated channel table.

3. The method for evaluating the signal-to-noise ratio of an image sensor based on a specific flat area according to claim 1, wherein: The step S5 further comprises the following steps: S501, calculating the mean of the grayscale values; S502, calculating the variance of the grayscale value; S503, calculating the standard deviation of the grayscale value; S504 , according to the results of steps S501 - S503 , calculate the quotient of the mean and the standard deviation to obtain the signal-to-noise ratio.

Images