KR20200141808A - Method and apparatus for detecting dead pixel - Google Patents

Abstract

According to an embodiment of the present invention, provided is a method for correcting non-uniformity based on a dead pixel detection scene. According to an embodiment of the present invention, local image data is aligned using a brightness value or a gain value, and a differential alignment threshold value is applied through differentiation to obtain a dead pixel classification criterion based on a local image, thereby detecting a dead pixel.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention Method and apparatus for detecting dead pixel

The present invention relates to a method and apparatus for detecting dead pixels.

Due to the characteristics of image sensors used for image detection such as infrared cameras and visible light cameras, non-uniformity and dead pixels occur in the focal plane array (FPA) due to the semiconductor manufacturing process. Currently, post-processing imaging techniques are mainly used to automatically correct dead pixels generated in the image sensor semiconductor process.

In the past, to detect dead pixels that occur in infrared cameras and visible light cameras, temperature resolution (Noise Equivalent Temperature Difference, hereinafter referred to as NETD), gain, and brightness values are applied globally to all pixels to cause dead. It mainly uses a method of detecting pixels.

However, since this method applies the same conditions to all pixels of an image, there is a limit to detecting most of the dead pixels. Even if a dead pixel is automatically detected while changing the standard for detecting a dead pixel finely, a phenomenon in which a plurality of dead pixels are simultaneously erroneously detected or not detected due to a difference in the minute reference value occurs.

[Prior technical literature number]

Prior 1: Korean Patent Registration No. 10-1673447

Prior 2: Korean Patent Registration No. 10-1958472

An embodiment is a dead pixel detection method and apparatus capable of detecting a dead pixel by obtaining a criterion for classifying a dead pixel based on a local image by aligning and differentiating local image data using a brightness value or a gain value and applying a differential alignment threshold value Is to provide.

A dead pixel detection method according to an embodiment includes the steps of: generating and aligning a predetermined array of local image data obtained from an input image, and differentiating the aligned array; Obtaining a criterion for classifying dead pixels in the array based on a predetermined differential array threshold value for the values of the differentiated array; And detecting a dead pixel in the local image data based on the acquired dead pixel classification criterion.

The dead pixel detection method may further include marking the detected dead pixel on the entire image data corresponding to the local image data.

The dead pixel is detected while moving a detection window corresponding to a unit of the local image data with respect to the entire image data.

The differential arrangement threshold value is characterized by a value obtained by multiplying a temporal noise by a predetermined constant α.

The differentiating step may include generating the obtained local image data in a one-dimensional array; Generating the array in which the generated one-dimensional array is arranged in ascending order; And generating a differential arrangement obtained by differentiating the image data values of the generated arrangement.

In the detecting step, a pixel corresponding to a position of the one-dimensional array having a value larger than the differential arrangement threshold value in the differential arrangement is detected as the dead pixel.

According to another exemplary embodiment, an apparatus for detecting a dead pixel may include an array alignment unit that generates and aligns a predetermined array of local image data acquired from an input image and differentiates the aligned array; A reference setting unit for obtaining a criterion for classifying dead pixels in the array based on a predetermined differential array threshold value for the values of the differentiated array; And a dead pixel detector configured to detect dead pixels in the local image data based on the acquired dead pixel classification criteria.

The dead pixel detection apparatus is characterized in that the detected dead pixel is marked on the entire image data corresponding to the local image data.

The dead pixel detection apparatus is characterized in that the dead pixel detection is performed while moving a detection window corresponding to a unit of the local image data with respect to the entire image data.

The differential arrangement threshold value is characterized by a value obtained by multiplying a temporal noise by a predetermined constant α.

The array alignment unit generates the obtained local image data as a one-dimensional array, generates the array in which the generated one-dimensional array is arranged in ascending order, and differentiates the image data values of the generated array. It is characterized by generating.

The dead pixel detection unit may detect a pixel corresponding to a position of the one-dimensional array having a value greater than the differential arrangement threshold value in the differential arrangement as the dead pixel.

A dead pixel detection apparatus according to another embodiment includes an image acquisition unit that acquires an input image; And a processor that detects at least one dead pixel from the acquired input image,

The processor generates and aligns a predetermined array of local image data from the input image, differentiates the sorted array, and calculates values of the differentiated array based on a predetermined differential array threshold. A dead pixel classification criterion is obtained, and a dead pixel in the local image data is detected based on the obtained dead pixel classification criterion.

The processor marks the detected dead pixel on the entire image data corresponding to the local image data, and detects the dead pixel while moving the detection window corresponding to the unit of the local image data with respect to the entire image data. Characterized in that.

And a recording medium on which a program for executing the dead pixel detection method according to another embodiment on a computer is recorded.

The dead pixel detection method and apparatus according to the embodiment may detect most of the dead pixels with a higher probability than a conventional method in which the same criterion is applied for dead pixel detection in all image pixels.

In addition, the dead pixel detection method and apparatus according to the embodiment is very less sensitive than the existing threshold values such as NETD, a gain value, and a brightness value by using a differential arrangement threshold value. Therefore, it is easy to set and use a threshold value with these advantages, so that mass production can be secured during mass production. Existing, when dead pixels are detected by applying the same NETD, Gain, and brightness values, it is possible to solve the disadvantage of large amounts of false or undetected dead pixels even with a small difference in reference value.

In addition, various types of image data, such as brightness, gain, and pixel NETD, can be applied and used.

1 is a schematic diagram of a dead pixel detection apparatus 100 according to an exemplary embodiment.
2 is a schematic diagram illustrating a dead pixel detection method according to another exemplary embodiment.
3A to 3C are exemplary diagrams for explaining distribution of local image data.
4 is a flow chart illustrating a dead pixel detection method according to another exemplary embodiment.
5 and 6 are exemplary diagrams for explaining a dead pixel detection result according to another embodiment.

The terms used in the embodiments have been selected as currently widely used general terms as possible while considering functions in the embodiments, but this may vary depending on the intention or precedent of a technician working in the art, the emergence of new technologies, etc. . In addition, in certain cases, there are terms that are arbitrarily selected, and in this case, the meaning will be described in detail in the description of the corresponding embodiment. Therefore, the terms used in the present embodiments should be defined based on the meaning of the term and the contents of the present embodiments, not a simple name of the term.

In the description of the embodiments, when a certain part is connected to another part, this includes not only a case in which it is directly connected, but also a case in which it is electrically connected with another component interposed therebetween. In addition, when a certain part includes a certain component, it means that other components may be further included rather than excluding other components unless specifically stated to the contrary. In addition, the term "... unit" described in the embodiments means a unit that processes at least one function or operation, which may be implemented as hardware or software, or a combination of hardware and software.

Terms such as “consisting of” or “comprising” used in the present embodiments should not be construed as necessarily including all of the various constituent elements or various steps described in the specification, and some constituent elements or some It should be construed that the steps may not be included, or may further include additional components or steps.

The description of the following embodiments should not be construed as limiting the scope of the rights, and what those skilled in the art can easily infer should be construed as belonging to the scope of the rights of the embodiments. Hereinafter, embodiments for illustration only will be described in detail with reference to the accompanying drawings.

In general, infrared cameras include dead pixels in which output values of specific pixels are abnormally displayed according to material characteristics of the infrared detector, semiconductor manufacturing process, temperature and environment, or exposure time of the infrared detector. Here, the dead pixel may be divided into a hard dead pixel having an abnormal physical characteristic of the pixel itself and a soft dead pixel having an abnormal output value during operation of the infrared detector. In an embodiment, the term “dead pixel” may include both meanings.

Hereinafter, exemplary embodiments will be described in detail with reference to the accompanying drawings.

1 is a schematic diagram of a dead pixel detection apparatus 100 according to an exemplary embodiment.

Referring to FIG. 1, the dead pixel detection apparatus 100 includes an image data acquisition unit 110, an array alignment unit 120, a reference setting unit 130, and a dead pixel detection unit 140. In an embodiment, the dead cell detection apparatus 100 generates and arranges a predetermined array of local image data from an input image, differentiates the aligned array, and determines the values of the differentiated array based on a predetermined differential alignment threshold value. A dead pixel classification criterion in the array is obtained, and a dead pixel in the local image data is detected based on the obtained dead pixel classification criterion. Here, it is described that the dead pixel detection apparatus 100 includes the image data acquisition unit 110, but it goes without saying that the image data acquisition unit 110 may be separated.

The image data acquisition unit 110 acquires an input image. Here, the input image may be, for example, an infrared image or a visible light image taken from an IR camera or a visible light camera, but is not limited thereto. An IR camera can take an infrared image using an infrared keyboard or a charge-coupled device (CCD) sensitive to infrared rays.

The array aligner 120 generates and aligns a predetermined array of local image data from the input image provided from the image data acquisition unit 110, and differentiates the sorted array. The input image can be divided into a predetermined size, for example, 16*16 size. The array alignment unit 120 generates local image data in a one-dimensional array. Then, after sorting the one-dimensional arrayed data values in ascending or descending order in order of size, a differential array is generated.

The reference setting unit 130 obtains a criterion for classifying dead pixels in an array arranged in order of size based on a predetermined differential arrangement threshold value for the values of the differential arrangement. Here, the differential arrangement threshold value is temporal noise. Noise) may be multiplied by a predetermined constant α, and the constant α may be an arbitrary constant, and may be set differently according to the performance and use of the dead pixel detection device.

The dead pixel detection unit 140 detects a dead pixel in local image data based on the dead pixel classification criterion obtained by the reference setting unit 130. The dead pixel detector 140 detects, as a dead pixel, a pixel corresponding to a position of a one-dimensional array having a value greater than the differential arrangement threshold value in the differential arrangement.

In an embodiment, the dead pixel detection apparatus 100 marks the detected dead pixel on the entire image data corresponding to the local image data. In addition, dead pixels may be detected while moving a detection window corresponding to a unit of local image data for all image data.

2 is a schematic diagram illustrating a dead pixel detection method according to another exemplary embodiment.

With reference to FIG. 2, a dead pixel detection method using a differential arrangement alignment technique will be described. Local image data 102 is acquired in a partial area from the entire image 101 obtained from the image sensor. Here, the local image data 102 is shown to have a size of 16*16, but the present invention is not limited thereto, and may be variously set according to the size of the window.

Then, the acquired local image data is generated in a one-dimensional array (103). At this time, the simple array of one-dimensional local image data does not have special features such as white noise. However, when the local image data is sorted 104 in ascending order, it has several characteristics capable of detecting dead pixels. First, very low or high value image data are located at both ends of the array, making it easier to find the dead pixel reference value. When the differential array 105 is obtained by differentiating based on the aligned one-dimensional local image data, it is possible to easily find a dead pixel reference value based on the peak value position of the differential array. Starting from the middle of the differential array, if a position of a one-dimensional array having a value higher than the differential array threshold value is found in two directions in which the image data value increases and the image data value decreases, the position of the alignment array 104 is based on this position. If you set a dead pixel reference value based on the corresponding image data value, you can quickly and easily find it based on the dead pixel. A dead pixel detection process is performed by detecting (108) a dead pixel in a local image using a reference value for classifying a dead pixel, and marking (109) a dead pixel position on the entire image based on this. Through this series of processes, a dead pixel may be detected in the entire image by scanning the entire image in the form of a sliding window at a window location for acquiring local image data.

3A to 3C are exemplary diagrams for explaining distribution of local image data.

3A to 3C, an alignment technique is applied to detect dead pixels in a local image. As shown in FIG. 3A, an image data array generated simply by using local image data and brightness value distribution has no characteristics like general noise data (201). However, as illustrated in FIG. 3B, when image data is aligned, data that can be classified as dead pixels are located at the front and rear portions of the alignment array (202). As shown in FIG. 3C, when a differential array is generated by differentiating the alignment array, a large peak value can be identified between the normal image data and the image data corresponding to the dead pixel (203). A reference value capable of distinguishing the dead pixels in the alignment arrangement 202 based on the arrangement position at which the peak value appears. When detecting a position at which the peak value appears in the differential array 203, the differential array threshold value may be used by multiplying the threshold value by a specific α value based on Temporal Noise (TN) (α x TN). In an embodiment, the differential arrangement threshold value is very insensitive compared to using a general reference pixel NETD, a brightness value, a gain value, and the like, and thus it is not necessary to determine it through a precise process, and a fixed value may be used. In addition, this insensitive differential arrangement threshold has an advantage in mass production as it can reduce a lot of tuning factors during mass production of products.

4 is a flow chart illustrating a dead pixel detection method according to another exemplary embodiment.

Referring to FIG. 4, in step 400, local image data is acquired from all image data. In step 402, an array of 1D image data is generated using the acquired local image data. In step 404, the generated 1D local image data array is arranged. In step 406, the ordered one-dimensional array is differentiated to generate a differential array. In step 408, an array position that exceeds the differential array threshold value in the differential array, starting from the middle of the array based on the differential array threshold value, is obtained in the forward and backward directions of the array. In step 410, a dead pixel classification reference image data value, for example, a brightness value or a gain value, is acquired from the alignment array based on the array position acquired in step 408. In step 412, a dead pixel is detected in the local image using the dead pixel classification criterion acquired in the previous process. In step 414, a dead pixel is marked in the entire image based on the position of the dead pixel detected in the local image. In step 416, the coordinates of the local image are moved, and steps 400 to 414 are performed. Through this process, dead pixels are detected on the entire image.

5 and 6 are exemplary diagrams for explaining a dead pixel detection result according to another embodiment.

An example of a result of detecting a local dead pixel applying the alignment technique will be described with reference to FIGS. 5 and 6. A result of detecting a dead pixel using the method suggested by the present invention using the same image sensor and the same image data is shown. As shown in FIG. 5, it can be seen that most of the dead pixels are detected when brightness data is used (401). And, as shown in FIG. 6, it can be confirmed that most of the dead pixels are stably detected even when the gain data is used (402). As described above, even in the two cases using the brightness value or the gain value of the image data, a stable result of detecting most of the dead pixels can be confirmed in the same manner.

The apparatus for detecting a dead pixel according to an embodiment applies the same differential alignment threshold value to local image data using a brightness value or a gain value using alignment and differentiation. In this case, a variable brightness value or a gain value may be obtained for classifying a dead pixel based on a local image. Dead pixels are detected using a brightness value or a gain value for classifying dead pixels obtained based on a local image.

The dead pixel detection method and apparatus according to the embodiment may detect most of the dead pixels with a higher probability than a conventional method in which the same criterion is applied for dead pixel detection in all image pixels. In addition, the dead pixel detection method and apparatus according to the embodiment is very less sensitive than the existing threshold values such as NETD, a gain value, and a brightness value by using a differential arrangement threshold value. Therefore, it is easy to set and use a threshold value with these advantages, so that mass production can be secured during mass production. Existing, when dead pixels are detected by applying the same NETD, Gain, and brightness values, it is possible to solve the disadvantage of large amounts of false or undetected dead pixels even with a small difference in reference value. In addition, various types of image data, such as brightness, gain, and pixel NETD, can be applied and used.

An embodiment of the present invention may also be implemented in the form of a recording medium including instructions executable by a computer, such as a program module executed by a computer. Computer-readable media can be any available media that can be accessed by a computer, and includes both volatile and nonvolatile media, removable and non-removable media. Further, the computer-readable medium may include both computer storage media and communication media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Communication media typically includes computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave, or other transmission mechanism, and includes any information delivery medium.

The above description of the present invention is for illustrative purposes only, and those of ordinary skill in the art to which the present invention pertains will be able to understand that other specific forms can be easily modified without changing the technical spirit or essential features of the present invention will be. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as being distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention. do.

Claims (15)

Generating and aligning a predetermined array of local image data obtained from an input image, and differentiating the aligned array;
Obtaining a criterion for classifying dead pixels in the array based on a predetermined differential array threshold value for the values of the differentiated array; And
And detecting a dead pixel in the local image data based on the acquired dead pixel classification criterion. The method of claim 1,
And marking the detected dead pixel on the entire image data corresponding to the local image data. The method of claim 2,
And detecting the dead pixel while moving a detection window corresponding to a unit of the local image data with respect to the entire image data. The method of claim 1,
The differential sequence threshold is,
A dead pixel detection method, characterized in that it is a value obtained by multiplying temporal noise by a predetermined constant α. The method of claim 1,
The differentiation step,
Generating the acquired local image data in a one-dimensional array;
Generating the array in which the generated one-dimensional array is arranged in ascending order; And
And generating a differential array obtained by differentiating the image data values of the generated array. The method of claim 5,
The detecting step,
And detecting a pixel corresponding to a position of the one-dimensional array having a value greater than the differential arrangement threshold value in the differential arrangement as the dead pixel. A recording medium storing a program for executing the method according to any one of claims 1 to 6 on a computer. An array aligner for generating and aligning a predetermined array of local image data obtained from an input image and differentiating the sorted array;
A reference setting unit for obtaining a criterion for classifying dead pixels in the array based on a predetermined differential array threshold value for the values of the differentiated array; And
A dead pixel detection apparatus comprising a dead pixel detection unit configured to detect a dead pixel in the local image data based on the acquired dead pixel classification criterion. The method of claim 8,
And marking the detected dead pixels on all image data corresponding to the local image data. The method of claim 9,
And detecting the dead pixel while moving a detection window corresponding to a unit of the local image data with respect to the entire image data. The method of claim 8,
The differential sequence threshold is,
A dead pixel detection device, characterized in that a value obtained by multiplying a temporal noise by a predetermined constant α. The method of claim 8,
The arrangement alignment unit,
Generating the obtained local image data as a one-dimensional array, generating the array in which the generated one-dimensional array is arranged in ascending order, and generating a differential array obtained by differentiating the image data values of the generated array. Dead pixel detection device. The method of claim 12,
The dead pixel detection unit,
And detecting a pixel corresponding to a position of the one-dimensional array having a value greater than the threshold value of the differential arrangement in the differential arrangement as the dead pixel. An image acquisition unit that acquires an input image;
And a processor that detects at least one dead pixel from the acquired input image,
The processor,
Create and align a predetermined array of local image data from the input image, differentiate the aligned array, and classify dead pixels in the array based on a predetermined differential alignment threshold value for the values of the differentiated array And detecting a dead pixel in the local image data based on the acquired dead pixel classification criterion. The method of claim 14,
The processor,
Marking the detected dead pixel on all image data corresponding to the local image data, and performing the dead pixel detection while moving a detection window corresponding to a unit of the local image data with respect to the entire image data. Dead pixel detection device.

Images