[go: up one dir, main page]

CN105554484B - A kind of demosaicing methods for being applied to high photosensitive micro- inclined array image-forming - Google Patents

A kind of demosaicing methods for being applied to high photosensitive micro- inclined array image-forming Download PDF

Info

Publication number
CN105554484B
CN105554484B CN201510883829.5A CN201510883829A CN105554484B CN 105554484 B CN105554484 B CN 105554484B CN 201510883829 A CN201510883829 A CN 201510883829A CN 105554484 B CN105554484 B CN 105554484B
Authority
CN
China
Prior art keywords
resolution
polarization
image
unbiased
pixel
Prior art date
Legal status (The legal status 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 status listed.)
Active
Application number
CN201510883829.5A
Other languages
Chinese (zh)
Other versions
CN105554484A (en
Inventor
赵永强
刘吾腾
李琳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Northwestern Polytechnical University
Original Assignee
Northwestern Polytechnical University
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 Northwestern Polytechnical University filed Critical Northwestern Polytechnical University
Priority to CN201510883829.5A priority Critical patent/CN105554484B/en
Publication of CN105554484A publication Critical patent/CN105554484A/en
Application granted granted Critical
Publication of CN105554484B publication Critical patent/CN105554484B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/10Cameras or camera modules comprising electronic image sensors; Control thereof for generating image signals from different wavelengths
    • GPHYSICS
    • G02OPTICS
    • G02BOPTICAL ELEMENTS, SYSTEMS OR APPARATUS
    • G02B27/00Optical systems or apparatus not provided for by any of the groups G02B1/00 - G02B26/00, G02B30/00
    • G02B27/28Optical systems or apparatus not provided for by any of the groups G02B1/00 - G02B26/00, G02B30/00 for polarising
    • G02B27/283Optical systems or apparatus not provided for by any of the groups G02B1/00 - G02B26/00, G02B30/00 for polarising used for beam splitting or combining
    • G02B27/285Optical systems or apparatus not provided for by any of the groups G02B1/00 - G02B26/00, G02B30/00 for polarising used for beam splitting or combining comprising arrays of elements, e.g. microprisms

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Optics & Photonics (AREA)
  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Transforming Light Signals Into Electric Signals (AREA)

Abstract

本发明公开了一种应用于高感光微偏阵列成像的去马赛克方法,将高感光微偏阵列固定设置于图像传感器靠近入射光的一侧,图像传感器将得到的图像数据发送至数字信号处理器,并按照以下步骤进行对图像数据去马赛克处理:步骤1、得到一幅低分辨率部分偏振角度图像;步骤2、得到高分辨率无偏图像,再得到一幅低分辨率无偏图像;步骤3、得到四幅不同偏振角度的中间图像,然后再减去低分辨率无偏图像,得到四副低分辨率偏振角度差图像;步骤4、得到一幅高分辨率角度差图像;步骤5、最终得到四副不同偏振角度的高分辨率图像,即完成原始图像的去马赛克处理,解决了高感光微偏阵列的去马赛克问题。

The invention discloses a demosaic method applied to high-sensitivity micro-polarization array imaging. The high-sensitivity micro-polarization array is fixedly arranged on the side of the image sensor close to the incident light, and the image sensor sends the obtained image data to a digital signal processor. , and perform image data demosaic processing according to the following steps: step 1, obtain a low-resolution partial polarization angle image; step 2, obtain a high-resolution unbiased image, and then obtain a low-resolution unbiased image; step 3. Obtain four intermediate images of different polarization angles, and then subtract the low-resolution unbiased images to obtain four low-resolution polarization angle difference images; step 4, obtain a high-resolution angle difference image; step 5, finally Four high-resolution images with different polarization angles are obtained, that is, the demosaicing process of the original image is completed, and the demosaicing problem of the high-sensitivity micro-polarization array is solved.

Description

一种应用于高感光微偏阵列成像的去马赛克方法A Demosaicing Method Applied to High Sensitivity Micro-polarized Array Imaging

技术领域technical field

本发明属于偏振光学成像及图像处理技术领域,涉及一种应用于高感光微偏阵列成像的去马赛克方法。The invention belongs to the technical field of polarization optical imaging and image processing, and relates to a demosaic method applied to high-sensitivity micro-polarization array imaging.

背景技术Background technique

由于高感光微偏阵列直接获取的图像中每个像素位置只对应一个偏振角度或者无偏的强度值,而为了获得场景的偏振信息,必须同时获得0°、45°、90°和135°四个偏振角度的图像,因此必须根据已经获得的像素强度信息通过去马赛克方法恢复出高分辨率的四幅偏振角度图像,而现有的去马赛克方法只适用于传统排布模式的微偏阵列,不能直接应用于高感光微偏阵列上面。Since each pixel position in the image directly acquired by the high-sensitivity micro-polarization array corresponds to only one polarization angle or an unbiased intensity value, in order to obtain the polarization information of the scene, it is necessary to obtain the 0°, 45°, 90° and 135° four Therefore, it is necessary to restore four high-resolution polarization angle images through the demosaic method according to the obtained pixel intensity information, and the existing demosaic method is only suitable for the slightly offset array of the traditional arrangement mode, and cannot It is directly applied on the high-sensitivity micro-polarization array.

发明内容Contents of the invention

本发明的目的是提供一种应用于高感光微偏阵列成像的去马赛克方法,以解决高感光微偏阵列的去马赛克问题。The object of the present invention is to provide a demosaicing method applied to the imaging of high-sensitivity micro-polarization arrays, so as to solve the demosaicing problem of high-sensitivity micro-polarization arrays.

本发明所采用的技术方案是,一种应用于高感光微偏阵列的去马赛克方法,其特征在于,将高感光微偏阵列固定设置于图像传感器靠近入射光的一侧,图像传感器将得到的图像数据发送至数字信号处理器,并按照以下步骤进行对图像数据去马赛克处理:The technical solution adopted in the present invention is a demosaic method applied to a high-sensitivity micro-polarization array, which is characterized in that the high-sensitivity micro-polarization array is fixedly arranged on the side of the image sensor close to the incident light, and the image sensor will obtain The image data is sent to the digital signal processor, and the image data is demosaiced according to the following steps:

步骤1、对于通过高感光微偏阵列和图像传感器采集到的图像数据,将其中各个偏振角度单元的偏振角度数据分别融合成一个值,得到一幅低分辨率部分偏振角度图像;Step 1. For the image data collected by the high-sensitivity micro-polarization array and the image sensor, the polarization angle data of each polarization angle unit are respectively fused into one value to obtain a low-resolution partial polarization angle image;

步骤2、估计高感光微偏阵列中偏振角度位置处像素的无偏强度数据,得到高分辨率无偏图像;Step 2. Estimating the unbiased intensity data of the pixel at the polarization angle position in the high-sensitivity micro-polarization array to obtain a high-resolution unbiased image;

再将高分辨率无偏图像的各个偏振角度单元中,通过估计出来的无偏强度数据求平均得到一个值,作为对应偏振角度单元的无偏强度数值,对整个阵列范围内进行同样的处理,即得到一幅低分辨率无偏图像;Then average the estimated unbiased intensity data in each polarization angle unit of the high-resolution unbiased image to obtain a value as the unbiased intensity value of the corresponding polarization angle unit, and perform the same processing on the entire array range. That is, a low-resolution unbiased image is obtained;

步骤3、将经步骤1得到的低分辨率部分偏振角度图像,在经步骤2处理得到的低分辨率无偏图像的指导下,通过插值得到四幅不同偏振角度的中间图像,然后再在得到的四个低分辨率部分偏振角度图中减去低分辨率无偏图像,即得到四副低分辨率偏振角度差图像;Step 3. Under the guidance of the low-resolution unbiased image obtained through step 2, the low-resolution partial polarization angle image obtained in step 1 is obtained by interpolation to obtain four intermediate images of different polarization angles, and then the obtained Subtract the low-resolution unbiased images from the four low-resolution partial polarization angle images to obtain four low-resolution polarization angle difference images;

步骤4、将经步骤3得到的四副低分辨率偏振角度差图像,经过双线性插值、上采样的处理方法,得到一幅高分辨率角度差图像;Step 4. The four low-resolution polarization angle difference images obtained in step 3 are processed by bilinear interpolation and upsampling to obtain a high-resolution angle difference image;

步骤5、对经步骤4得到的高分辨率角度差图像与经步骤2得到的高分辨率无偏图像求和,最终得到四副不同偏振角度的高分辨率图像,即完成原始图像的去马赛克处理;Step 5. Sum the high-resolution angular difference image obtained in step 4 and the high-resolution unbiased image obtained in step 2, and finally obtain four high-resolution images with different polarization angles, that is, complete the demosaicing of the original image deal with;

其中,高感光微偏阵列包括依次紧密排列的若干个最小周期单元,每个最小周期单元均包括四个不同偏振角度的偏振单元,四个不同偏振角度分别为0°、45°、90°或135°,各个偏振单元的镜片包括若干个偏振片和无偏镜片。Among them, the high-sensitivity micro-polarization array includes several minimum periodic units closely arranged in sequence, and each minimum periodic unit includes four polarization units with different polarization angles, and the four different polarization angles are 0°, 45°, 90° or 135°, the lenses of each polarization unit include several polarizers and non-polarizers.

进一步的,在每个偏振单元内的无偏镜片和偏振片为交错排列设置。Further, the non-polarized lenses and polarizers in each polarizing unit are arranged in a staggered arrangement.

进一步的,无偏镜片或偏振片,分别沿水平、竖直或对角线的方向连续排列。Further, the non-polarized lenses or polarizers are respectively arranged continuously along the horizontal, vertical or diagonal directions.

进一步的,步骤1中,将各个偏振角度单元的偏振角度数据,分别通过求平均值的方法融合成一个值。Further, in step 1, the polarization angle data of each polarization angle unit are respectively fused into one value by means of averaging.

进一步的,步骤2中,低分辨率无偏图像与通过步骤1得到的低分辨率部分偏振角度图像的大小相同。Further, in step 2, the size of the low-resolution unbiased image is the same as that of the low-resolution partial polarization angle image obtained in step 1.

进一步的,步骤3中,中间图像中某一待求像素的估计方法为:根据待求像素的邻域中四个与之对应相同角度偏振片的已知像素值,来估计待求像素的数值,其中,待求像素分为两类,第一类待求像素,其对应的偏振角度与其主副对角线相邻像素相同,第二类待求像素,其对应的偏振角度与其水平或者垂直方向的相邻像素相同。Further, in step 3, the method of estimating a certain pixel to be sought in the intermediate image is: to estimate the value of the pixel to be sought according to the known pixel values of four corresponding polarizers of the same angle in the neighborhood of the pixel to be sought , where the pixels to be requested are divided into two categories, the first type of pixels to be requested, whose corresponding polarization angle is the same as that of the adjacent pixels on the main and secondary diagonals, and the second type of pixels to be requested, whose corresponding polarization angle is horizontal or vertical to Neighboring pixels in the same direction.

进一步的,第二类待求像素的估计方法为,首先估计第一类像素位置处像素的值,将第一类像素位置处像素的估计值作为该位置的已知值,来估计第二类像素位置处像素的值。Further, the method for estimating the second type of pixel to be sought is to first estimate the value of the pixel at the position of the first type of pixel, and use the estimated value of the pixel at the position of the first type of pixel as the known value of the position to estimate the value of the pixel of the second type The value of the pixel at the pixel location.

进一步的,步骤4中,高分辨率角度差图像与通过步骤2中得到的高分辨率无偏图像的大小相同。Further, in step 4, the high-resolution angular difference image has the same size as the high-resolution unbiased image obtained in step 2.

本发明的有益效果是,针对高感光微偏阵列微偏阵列设计的去马赛克方法,解决了现有去马赛克方法只适用于传统排布模式的微偏阵列,不能直接应用于高感光微偏阵列的问题;充分利用了高感光偏振阵列的特性,利用无偏镜片采集到的数据来估计恢复偏振数据时的所需要的梯度信息,同时又具备传统微偏阵列去马赛克的优点,即利用邻域内尽可能多的信息来估计缺失的数据。联合以上两种手段以尽可能地保证最终获得偏振信息的准确性。The beneficial effect of the present invention is that the demosaicing method designed for the micro-biased array of the highly photosensitive micro-polarized array solves the problem that the existing demosaicing method is only applicable to the micro-biased array of the traditional arrangement mode, and cannot be directly applied to the highly photosensitive micro-biased array The problem; make full use of the characteristics of the high-sensitivity polarization array, use the data collected by the unpolarized lens to estimate the gradient information required to restore the polarization data, and at the same time have the advantages of the traditional micro-polarization array demosaic, that is, use the neighborhood As much information as possible to estimate missing data. The above two means are combined to ensure the accuracy of the final polarization information as much as possible.

附图说明Description of drawings

图1为本发明设有高感光微偏阵列的数码相机系统;Fig. 1 is provided with the digital camera system of high-sensitivity micro-polarization array of the present invention;

其中,1.入射光,2.镜头,3.高感光微偏阵列,4.图像传感器,5.模拟信号处理器,6.D/A转换器,7.数字信号处理器,8.图像显示器;Among them, 1. Incident light, 2. Lens, 3. High-sensitivity micro-polarization array, 4. Image sensor, 5. Analog signal processor, 6. D/A converter, 7. Digital signal processor, 8. Image display ;

图2为数码相机系统中数字信号处理器的功能模块示意图;2 is a schematic diagram of functional modules of a digital signal processor in a digital camera system;

图中,11.低分辨率部分偏振角度模块,12.高分辨率无偏模块,13.低分辨率无偏模块,14.低分辨率偏振角度差模块,15.高分辨率偏振角度差模块,16.高分辨率最终图像模块;In the figure, 11. Low-resolution partial polarization angle module, 12. High-resolution unbiased module, 13. Low-resolution unbiased module, 14. Low-resolution polarization angle difference module, 15. High-resolution polarization angle difference module , 16. High resolution final image module;

图3为高感光微偏阵列的最小周期单元;Fig. 3 is the minimum periodic unit of the highly sensitive micro-polarization array;

图3中,A.0°偏振单元,B.45°偏振单元,C.90°偏振单元,D.135°偏振单元,E.0°偏振单元中的偏振片组;In Fig. 3, A.0 ° polarizing unit, B.45 ° polarizing unit, C.90 ° polarizing unit, D.135 ° polarizing unit, polarizer group in E.0 ° polarizing unit;

图4为输入低分辨率偏振角度差模块的图像已知信息的部分示意图。Fig. 4 is a partial schematic diagram of the known image information input into the low-resolution polarization angle difference module.

具体实施方式detailed description

下面结合附图和具体实施方式对本发明进行详细说明。The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

本发明提供了一种应用于高感光微偏阵列的去马赛克方法,将高感光微偏阵列固定设置于图像传感器靠近入射光的一侧,图像传感器将得到的图像数据发送至数字信号处理器7,并按照以下步骤进行对图像数据去马赛克处理:The present invention provides a demosaic method applied to a high-sensitivity micro-biased array. The high-sensitivity micro-slanted array is fixedly arranged on the side of the image sensor close to the incident light, and the image sensor sends the obtained image data to the digital signal processor 7 , and follow the steps below to demosaic the image data:

步骤1、对于图像传感器采集到的图像数据,将其中各个偏振角度单元的偏振角度数据分别融合成一个值,得到一幅低分辨率部分偏振角度图像。Step 1. For the image data collected by the image sensor, the polarization angle data of each polarization angle unit are respectively fused into one value to obtain a low-resolution partial polarization angle image.

步骤2、估计高感光微偏阵列中偏振角度位置处像素的无偏强度数据,得到一幅大小与原始图像相同的高分辨率无偏图像;Step 2. Estimate the unbiased intensity data of the pixel at the polarization angle position in the high-sensitivity micro-polarization array, and obtain a high-resolution unbiased image with the same size as the original image;

再将各个偏振角度单元中通过估计出来的无偏强度数据求平均得到一个值,作为对应偏振角度单元的无偏强度数值,对整个阵列范围内进行同样的处理,即得到一幅低分辨率无偏图像;所述低分辨率无偏图像与通过步骤1得到的低分辨率部分偏振角度图像的大小相同。Then average the estimated unbiased intensity data in each polarization angle unit to obtain a value, which is used as the unbiased intensity value of the corresponding polarization angle unit, and the same process is performed on the entire array range to obtain a low-resolution unbiased intensity value. polarized image; the low-resolution unbiased image has the same size as the low-resolution partial polarization angle image obtained through step 1.

步骤3、将经步骤1得到的低分辨率部分偏振角度图像,在经步骤2处理得到的低分辨率无偏图像的指导下,通过插值得到四幅不同偏振角度的中间图像,然后再在得到的四个角度图中减去低分辨率无偏图像,即得到低分辨率偏振角度差图像;Step 3. Under the guidance of the low-resolution unbiased image obtained through step 2, the low-resolution partial polarization angle image obtained in step 1 is obtained by interpolation to obtain four intermediate images of different polarization angles, and then the obtained Subtract the low-resolution unbiased image from the four angle images to obtain the low-resolution polarization angle difference image;

步骤4、将经步骤3得到的低分辨率偏振角度差图像,经过双线性插值、上采样的处理方法,得到一幅高分辨率角度差图像,所述高分辨率角度差图像与通过步骤2中得到的高分辨率无偏图像的大小相同。Step 4. The low-resolution polarization angle difference image obtained in step 3 is processed through bilinear interpolation and upsampling to obtain a high-resolution angle difference image. The high-resolution angle difference image and the step The high-resolution unbiased images obtained in 2 have the same size.

步骤5、对经步骤4得到的高分辨率角度差图像与经步骤2得到的高分辨率无偏图像求和,最终得到四副不同偏振角度的高分辨率图像,即完成原始图像的去马赛克处理。Step 5. Sum the high-resolution angular difference image obtained in step 4 and the high-resolution unbiased image obtained in step 2, and finally obtain four high-resolution images with different polarization angles, that is, complete the demosaicing of the original image deal with.

实施例:Example:

将高感光偏振阵列3设置于图像传感器4靠近入射光的一侧,图像传感器4将得到的图像数据发送至数字信号处理器7,并按照以下步骤进行对图像数据去马赛克处理:The high-sensitivity polarization array 3 is arranged on the side of the image sensor 4 close to the incident light, the image sensor 4 sends the obtained image data to the digital signal processor 7, and the image data is demosaiced according to the following steps:

1)假设高感光偏振阵列3的最小周期单元如图3所示,那么通过最小单元中的0°偏振角度单元A和图像传感器4后,采集到的数据包含八个无偏数据,同时还包含一个或者多个0°偏振角度数据。其中,0°偏振角度数据的采集数量是由采集原始数据的图像传感器4所选择的操作模式确定的。1) Assuming that the minimum periodic unit of the high-sensitivity polarization array 3 is shown in Figure 3, after passing through the 0° polarization angle unit A and the image sensor 4 in the minimum unit, the collected data contains eight unbiased data, and also contains One or more 0° polarization angle data. Wherein, the collection quantity of 0° polarization angle data is determined by the operation mode selected by the image sensor 4 that collects the raw data.

根据图1所示的相机的内部结构,上述采集到的无偏数据和偏振角度数据,会经过模拟信号处理器5和D/A转换器6处理后传送到数字信号处理器7中,记为信号a,整个去马赛克算法是在数字信号处理器7中进行的,数字信号处理器7中内部包含的功能模块如图2所示。According to the internal structure of the camera shown in Figure 1, the unbiased data and polarization angle data collected above will be processed by the analog signal processor 5 and the D/A converter 6 and then sent to the digital signal processor 7, denoted as For signal a, the entire demosaicing algorithm is performed in the digital signal processor 7, and the functional modules included in the digital signal processor 7 are shown in FIG. 2 .

如图2所示,处理过的信号a,同时传送给低分辨率部分偏振角度模块11和高分辨率无偏模块12中,低分辨率部分偏振角度模块11将每个偏振角度单元中所包含的n个偏振角度数据融合为一个值,可以通过求和取平均来得到。这样一来,如图3所示的最小周期单元中,每个偏振角度单元就只对应一个偏振角度值,而非八个偏振角度值,因此称经过低分辨率部分偏振角度模块11的得到基于偏振角度单元而非像素级别的图像为低分辨率部分偏振角度图像。As shown in Figure 2, the processed signal a is sent to the low-resolution partial polarization angle module 11 and the high-resolution unbiased module 12 at the same time, and the low-resolution partial polarization angle module 11 will be included in each polarization angle unit The n polarization angle data of are fused into one value, which can be obtained by summing and averaging. In this way, in the minimum periodic unit shown in Figure 3, each polarization angle unit corresponds to only one polarization angle value, rather than eight polarization angle values, so it is said that the obtained low-resolution partial polarization angle module 11 is based on Polarization angle unit rather than pixel level images are low resolution partial polarization angle images.

2)在图2中,数字信号处理器7中的高分辨率无偏模块12,用来估计高感光微偏阵列中偏振片位置处像素的无偏强度数据。2) In FIG. 2, the high-resolution unbiased module 12 in the digital signal processor 7 is used to estimate the unbiased intensity data of the pixel at the position of the polarizer in the high-sensitivity micro-polarized array.

以图3中0度偏振角度单元A为例,高分辨率无偏模块2能够估计出0°偏振单元中偏振片组E中各个像素位置的无偏数据值,以估计偏振片组E中第二行第二列,编号为22位置处的像素强度值I22为例,I22=(I12+I32)/2;其他位置的估计方法与之类似。当高分辨率无偏模块12处理完毕后,可得到一幅大小与高感光微偏阵列大小相同的高分辨率无偏图像。Taking the 0° polarization angle unit A in Figure 3 as an example, the high-resolution unbiased module 2 can estimate the unbiased data value of each pixel position in the polarizer group E in the 0° polarization unit, so as to estimate the The second column of the second row, numbered as the pixel intensity value I 22 at the position 22 is taken as an example, I 22 =(I 12 +I 32 )/2; the estimation method for other positions is similar. After the high-resolution unbiased module 12 finishes processing, a high-resolution unbiased image with the same size as the high-sensitivity micro-biased array can be obtained.

低分辨率无偏模块13,用于获得高分辨率无偏模块12中输出的高分辨率无偏图像,然后分别将图像中各个偏振角度单元位置内估计出来的无偏数据求和取平均得到一个值,作为对应偏振角度单元的无偏强度数据,这样每个偏振角度单元对应一个无偏强度数据,因此称处理后得到的图像为低分辨率无偏图像。该低分辨率无偏图像与通过低分辨率部分偏振角度模块11得到的低分辨率部分偏振角度图像的大小相同。The low-resolution unbiased module 13 is used to obtain the high-resolution unbiased image output in the high-resolution unbiased module 12, and then respectively sum and average the unbiased data estimated in each polarization angle unit position in the image to obtain A value, as the unbiased intensity data corresponding to the polarization angle unit, so that each polarization angle unit corresponds to an unbiased intensity data, so the image obtained after processing is called a low-resolution unbiased image. The low-resolution unbiased image has the same size as the low-resolution partial polarization angle image obtained by the low-resolution partial polarization angle module 11 .

3)低分辨率偏振角度差模块14,用于接收低分辨率部分偏振角度模块11和低分辨率无偏模块13的输出图像,然后,低分辨率部分偏振角度图像在低分辨率无偏图像的指导下通过插值得到四幅不同偏振角度的中间图像。3) The low-resolution polarization angle difference module 14 is used to receive the output images of the low-resolution partial polarization angle module 11 and the low-resolution unbiased module 13, and then, the low-resolution partial polarization angle image is added to the low-resolution unbiased image Under the guidance of , four intermediate images with different polarization angles were obtained by interpolation.

该中间图像的估计方法为:为了提高偏振角度中间图像估计的准确性,我们估计中间图像中某个像素的值时,需要用到该像素邻域中四个与之对应相同角度偏振片的已知像素值来估计,而按照高感光微偏阵列的排布方式,待求像素可分为两类,一类为待求像素对应的偏振角度与其主副对角线相邻像素相同,另一类为待求像素对应的偏振角度与其水平或者垂直方向的相邻像素相同。第一类情况满足我们提出的要求,而第二类情况只能通过单个方向上的两个像素值来估计,为了达到我们的要求,我们采取了以下解决方法:首先估计第一类像素位置处像素的值,求得估计值之后作为该位置的已知值来估计第二类像素位置处像素的值。The estimation method of the intermediate image is as follows: in order to improve the accuracy of the estimation of the intermediate image of the polarization angle, when we estimate the value of a certain pixel in the intermediate image, we need to use four corresponding polarizers with the same angle in the neighborhood of the pixel. According to the arrangement of the high-sensitivity micro-polarization array, the pixels to be obtained can be divided into two categories, one is that the polarization angle corresponding to the pixel to be obtained is the same as that of the adjacent pixels on the main and secondary diagonals, and the other is The category is that the polarization angle corresponding to the pixel to be obtained is the same as that of its adjacent pixels in the horizontal or vertical direction. The first type of situation meets our requirements, while the second type of situation can only be estimated by two pixel values in a single direction. In order to meet our requirements, we have adopted the following solution: first estimate the pixel position of the first type The value of the pixel, after obtaining the estimated value, it is used as the known value of the position to estimate the value of the pixel at the second type of pixel position.

以求0°偏振中间图像为例,如图4所示,需要求出非0°偏振角度单元中如22(代表第二行第二列位置处),23(代表第二行第三列位置处),24(代表第二行第四列位置处)位置处对应的0°值。图4中,每个方框代表低分辨率无偏图像和低分辨率部分偏振角度图像两幅图像相同位置处的像素,方框中的I代表该位置处无偏强度值,I0,I45,I135分别代表该位置处的0度,45度,135度偏振角度的值。Take the 0° polarized intermediate image as an example, as shown in Figure 4, it is necessary to obtain non-0° polarization angle units such as 22 (representing the position of the second row and the second column), 23 (representing the position of the second row and the third column) position), 24 (representing the position of the fourth column in the second row) corresponds to the 0° value at the position. In Fig. 4, each box represents the pixel at the same position of the two images of the low-resolution unbiased image and the low-resolution partial polarization angle image, and the I in the box represents the unbiased intensity value at this position, I0, I45, I135 respectively represent the polarization angle values of 0 degree, 45 degree and 135 degree at this position.

步骤1、以求编号为22处的0°值I022为例,具体方法如下:Step 1, taking the 0° value I0 22 at number 22 as an example, the specific method is as follows:

1.1)先计算主对角线方向(Lgrad)和副对角线方向的梯度(Cgrad),I代表无偏强度值:1.1) First calculate the gradient (Cgrad) in the main diagonal direction (Lgrad) and the subdiagonal direction (Cgrad), and I represents the unbiased intensity value:

Lgrad=|I11-I33|+|2*I22-I11-I33|,Lgrad=|I 11 -I 33 |+|2*I 22 -I 11 -I 33 |,

Cgrad=|I13-I31|+|2*I22-I13-I31|, Cgrad =|I13- I31 |+| 2 *I22- I13 - I31 |,

1.2)计算不同对角线方向上所求像素位置的像素值:1.2) Calculate the pixel value of the requested pixel position in different diagonal directions:

LI022=(I011+I033)/2+(2*I22-I11-I33)/2,LI0 22 = (I0 11 +I0 33 )/2+(2*I 22 -I 11 -I 33 )/2,

CI022=(I013+I031)/2+(2*I22-I13-I31)/2,CI0 22 = (I0 13 +I0 31 )/2+(2*I 22 -I 13 -I 31 )/2,

1.3)将第一步计算出的值与经验设定的阈值T比较,确定所求像素值:1.3) Compare the value calculated in the first step with the empirically set threshold T to determine the desired pixel value:

If MAX(Lgrad,Cgrad)<T,If MAX(Lgrad, Cgrad)<T,

I022=(LI022+CI022)/2,I0 22 = (LI0 22 +CI0 22 )/2,

elseif Lgrad<Cgrad,elseif Lgrad<Cgrad,

I022=LI022I0 22 = LI0 22 ,

Else,Else,

I022=CI022I0 22 =CI0 22 ,

End。End.

然后根据估计得到的值在非对角线方向,即水平和垂直方向,估计出相应缺失角度的值。Then, according to the estimated values in non-diagonal directions, that is, the horizontal and vertical directions, the values of the corresponding missing angles are estimated.

步骤2、以求第二行第三列编号为23处的0°值为例,具体方法如下:Step 2. Take the 0° value in the second row and third column numbered 23 as an example, the specific method is as follows:

2.1)先分别计算水平方向(Hgrad)和竖直方向(Vgrad)上的梯度:2.1) First calculate the gradients in the horizontal direction (Hgrad) and the vertical direction (Vgrad):

Hgrad=|I22-I24|+|2*I23-I22-I24|, Hgrad =| I22 -I24|+| 2 * I23 -I22- I24 |,

Vgrad=|I13-I33|+|2*I23-I13-I33|,Vgrad=|I 13 -I 33 |+|2*I 23 -I 13 -I 33 |,

2.2)分别计算水平竖直方向上所求像素位置的像素值:2.2) Calculate the pixel values of the requested pixel positions in the horizontal and vertical directions respectively:

HI023=(I022+I024)/2+(2*I23-I22-I24)/2,HI0 23 = (I0 22 +I0 24 )/2+(2*I 23 -I 22 -I 24 )/2,

VI023=(I013+I033)/2+(2*I23-I13-I33)/2,VI0 23 = (I0 13 +I0 33 )/2+(2*I 23 -I 13 -I 33 )/2,

2.3)将第一步计算出的值与经验设定的阈值T比较,确定所求像素值:2.3) Compare the value calculated in the first step with the empirically set threshold T to determine the desired pixel value:

If MAX(Hgrad,Vgrad)<T,If MAX(Hgrad, Vgrad)<T,

I023=(HI023+VI023)/2,I0 23 =(HI0 23 +VI0 23 )/2,

elseif Hgrad<Vgrad,elseif Hgrad<Vgrad,

I023=HI023I0 23 =HI0 23 ,

Else,Else,

I023=VI023I0 23 = VI0 23 ,

End。End.

这样可求得除阵列最外层位置的像素外所有其他像素位置的0°像素值,类似地我们可以求出另外三个角度45°,90°和135°的像素值。In this way, the 0° pixel values of all other pixel positions except the pixel at the outermost position of the array can be obtained. Similarly, we can obtain the pixel values of the other three angles of 45°, 90° and 135°.

其中,I是无偏强度值,I0、I45、I90和I135分别为0度、45度、90度和135度偏振角度值;无偏强度值I右下角的角标(xy)代表像素的位置,x代表像素所在行数,y代表像素所在列数;L和C分别代表主对角线和副对角线方向,H和V分别代表水平方向和竖直方向,所以LI022表示第二行第二列主对角线方向上估计出的0度偏振角度值。Among them, I is the unbiased intensity value, I0, I45, I90 and I135 are the polarization angle values of 0 degrees, 45 degrees, 90 degrees and 135 degrees respectively; the subscript (xy) in the lower right corner of the unbiased intensity value I represents the position of the pixel , x represents the number of rows where the pixel is located, y represents the number of columns where the pixel is located; L and C represent the main diagonal and sub-diagonal directions respectively, H and V represent the horizontal direction and vertical direction respectively, so LI0 22 represents the second row Estimated 0 degree polarization angle value in the direction of the main diagonal in the second column.

在得到了低分辨率部分偏振角度中间图像之后,再在得到的四幅不同偏振角度的中间图像中,减去低分辨率无偏图像,这样就得到了低分辨率偏振角度差图像。After the low-resolution intermediate images of partial polarization angles are obtained, the low-resolution unbiased images are subtracted from the obtained four intermediate images of different polarization angles, so that the low-resolution polarization angle difference images are obtained.

4)将低分辨率偏振角度差图像输入高分辨率偏振角度差模块15,经过双线性插值、上采样技术使其与高感光微偏阵列的大小相同,最终处理完毕后得到的高分辨率角度差图像大小与高分辨率无偏模块12中输出的图像相同。4) Input the low-resolution polarization angle difference image into the high-resolution polarization angle difference module 15, make it the same size as the high-sensitivity micro-polarization array through bilinear interpolation and up-sampling technology, and obtain the high-resolution image after final processing The size of the angle difference image is the same as the image output from the high-resolution unbiased module 12 .

5)高分辨率最终图像模块16的输入为高分辨率无偏模块12和高分辨率偏振角度差模块15输出的图像,通过对高分辨率角度差图像与高分辨率无偏图像求和,得到最终的高分辨率图像b,然后将高分辨率图像b传送到的图像显示器8中显示出来。5) The input of the high-resolution final image module 16 is the image output by the high-resolution unbiased module 12 and the high-resolution polarization angle difference module 15, by summing the high-resolution angle difference image and the high-resolution unbiased image, The final high-resolution image b is obtained, and then the high-resolution image b is sent to the image display 8 for display.

本发明针对高感光微偏阵列微偏阵列设计的去马赛克方法,解决了现有去马赛克方法只适用于传统排布模式的微偏阵列,不能直接应用于高感光微偏阵列的问题;充分利用了高感光偏振阵列的特性,利用无偏镜片采集到的数据来估计恢复偏振数据时的所需要的梯度信息,同时又具备传统微偏阵列去马赛克的优点,即利用邻域内尽可能多的信息来估计缺失的数据。联合以上两种手段以尽可能地保证最终获得偏振信息的准确性。The present invention aims at the demosaicing method designed for the highly photosensitive micro-biased array micro-biased array, which solves the problem that the existing demosaicing method is only applicable to the micro-biased array of the traditional arrangement mode, and cannot be directly applied to the highly photosensitive micro-biased array; fully utilizes The characteristics of the high-sensitivity polarization array are used, and the data collected by the unpolarized lens is used to estimate the gradient information required to restore the polarization data. At the same time, it has the advantages of the traditional micro-polarization array demosaic, that is, using as much information as possible in the neighborhood to estimate missing data. The above two means are combined to ensure the accuracy of the final polarization information as much as possible.

Claims (8)

1. a kind of demosaicing methods for being applied to high photosensitive micro- inclined array, it is characterised in that the photosensitive micro- inclined array (3) of height is solid Surely be arranged at imaging sensor (4) close to incident light (1) side, imaging sensor (4) by obtained view data send to Digital signal processor (7), and follow the steps below to the processing of view data demosaicing:
Step 1, the view data for being collected by high photosensitive micro- inclined array (3) and imaging sensor (4), will wherein each The polarization angle data of polarization angle unit are fused into a value respectively, obtain a width low resolution partial polarization angular image;
In step 2, the high photosensitive micro- inclined array of estimation at polarization angle position pixel unbiased intensity data, obtain high-resolution without Inclined image;
Again by each polarization angle unit of high-resolution unbiased graph picture, it is averaging by the unbiased intensity data estimated A value is obtained, as the unbiased strength values of correspondence polarization angle unit, to carrying out same processing in whole array range, Obtain a width low resolution unbiased graph picture;
Step 3, the low resolution partial polarization angular image that will be obtained through step 1, obtained low resolution is being handled through step 2 Under the guidance of unbiased graph picture, the intermediate image of four width different polarization angles is obtained by interpolation, then again in obtained fourth officer Between subtract low resolution unbiased graph picture in image, that is, obtain fourth officer low resolution polarization angle difference image;
Step 4, the fourth officer low resolution polarization angle difference image that will be obtained through step 3, by bilinear interpolation, the place of up-sampling Reason method, obtains a panel height resolution angle difference image;
Step 5, the high-resolution unbiased graph picture to the high-resolution angle difference image that is obtained through step 4 with being obtained through step 2 are asked With finally give the high-definition picture of different polarization angle, that is, complete the demosaicing processing of original image;
Wherein, the photosensitive micro- inclined array (3) of the height includes several compact arranged minimum period units successively, each described Minimum period unit includes the polarization unit of four different polarization angles, four different polarization angles are respectively 0 °, 45 °, 90 ° or 135 °, the eyeglass of the polarization unit described in each includes several polarizers (I) and without deviation lens (II).
2. demosaicing methods as claimed in claim 1, it is characterised in that in each polarization unit without deviation lens It is the setting that is staggered with polarizer.
3. demosaicing methods as claimed in claim 1, it is characterised in that described without deviation lens or polarizer, respectively along water Flat, vertical or cornerwise direction continuous arrangement.
4. the demosaicing methods as described in any one in claims 1 to 3, it is characterised in that in the step 1, will be each The polarization angle data of individual polarization angle unit, are fused into a value by the method averaged respectively.
5. the demosaicing methods as described in any one in claims 1 to 3, it is characterised in that described in the step 2 The size of low resolution partial polarization angular image of the low resolution unbiased graph picture with being obtained by step 1 is identical.
6. the demosaicing methods as described in any one in claims 1 to 3, it is characterised in that described in the step 3 The method of estimation of a certain pixel to be asked is in intermediate image:It is inclined that equal angular is corresponded to therewith according to four in the neighborhood of pixel to be asked Shake the known pixel values of piece, to estimate the numerical value of the pixel to be asked, wherein, the pixel to be asked is divided into two classes, and the first kind is treated Pixel is sought, its corresponding polarization angle is identical with its major-minor diagonal adjacent pixels, Equations of The Second Kind pixel to be asked, its corresponding polarization Angle is identical with the adjacent pixel in its horizontally or vertically direction.
7. demosaicing methods as claimed in claim 6, it is characterised in that the method for estimation of Equations of The Second Kind pixel to be asked is, first First estimate the value of first kind pixel position pixel to be asked, regard the estimate of first kind pixel position pixel to be asked as the position The given value put, to estimate the value of Equations of The Second Kind pixel position pixel to be asked.
8. the demosaicing methods as described in any one in claims 1 to 3, it is characterised in that described in the step 4 High-resolution angle difference image is identical with the size of the high-resolution unbiased graph picture by being obtained in step 2.
CN201510883829.5A 2015-12-05 2015-12-05 A kind of demosaicing methods for being applied to high photosensitive micro- inclined array image-forming Active CN105554484B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510883829.5A CN105554484B (en) 2015-12-05 2015-12-05 A kind of demosaicing methods for being applied to high photosensitive micro- inclined array image-forming

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510883829.5A CN105554484B (en) 2015-12-05 2015-12-05 A kind of demosaicing methods for being applied to high photosensitive micro- inclined array image-forming

Publications (2)

Publication Number Publication Date
CN105554484A CN105554484A (en) 2016-05-04
CN105554484B true CN105554484B (en) 2017-07-11

Family

ID=55833395

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510883829.5A Active CN105554484B (en) 2015-12-05 2015-12-05 A kind of demosaicing methods for being applied to high photosensitive micro- inclined array image-forming

Country Status (1)

Country Link
CN (1) CN105554484B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111664943B (en) * 2020-04-29 2021-07-02 西北工业大学 Fabrication method of split-focal plane polarization detector and orthogonal polarization image acquisition method
CN112788314B (en) * 2021-01-20 2022-05-31 西北工业大学深圳研究院 Polarization demosaicing method based on edge change induction
CN114842046A (en) * 2022-05-09 2022-08-02 北京环境特性研究所 Device and method for measuring high-dynamic scene image of moving object
CN119251098B (en) * 2024-12-05 2025-03-21 长春理工大学 A highly robust demosaicing bilinear interpolation method for polarization hyperspectral imaging

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102749720A (en) * 2012-07-24 2012-10-24 哈尔滨工业大学 Depolarization beam splitting system based on prisms and bandpass filters
CN103293756A (en) * 2013-05-30 2013-09-11 京东方科技集团股份有限公司 Display processing device, system and method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7760256B2 (en) * 2007-05-31 2010-07-20 Panasonic Corporation Image processing apparatus that obtains color and polarization information

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102749720A (en) * 2012-07-24 2012-10-24 哈尔滨工业大学 Depolarization beam splitting system based on prisms and bandpass filters
CN103293756A (en) * 2013-05-30 2013-09-11 京东方科技集团股份有限公司 Display processing device, system and method

Also Published As

Publication number Publication date
CN105554484A (en) 2016-05-04

Similar Documents

Publication Publication Date Title
JP4019417B2 (en) Image processing apparatus and method, recording medium, and program
CN111225135B (en) Image sensor, imaging device, electronic equipment, image processing system and signal processing method
CN105554484B (en) A kind of demosaicing methods for being applied to high photosensitive micro- inclined array image-forming
CN103444183B (en) Color imaging element and imaging device
JP5128726B1 (en) Solid-state imaging device and imaging apparatus including the device
CN105445952B (en) The micro- inclined array and its imaging device of high photosensitive real-time polarization imaging
CN108305233B (en) A light field image correction method for microlens array errors
JP2010057067A (en) Image pickup apparatus and image processing apparatus
JP2013223209A (en) Image pickup processing device
CN101198965A (en) Correction of Defective Pixels in Image Sensors
US20150116562A1 (en) Compact spacer in multi-lens array module
JP2022046601A (en) Plenoptic sub aperture view shuffling with improved resolution
US20100253817A1 (en) Orientation-based approach for forming a demosaiced image, and for color correcting and zooming the demosaiced image
Gao et al. Image interpolation methods evaluation for division of focal plane polarimeters
Gao et al. Gradient based interpolation for division of focal plane polarization imaging sensors
KR102437510B1 (en) Plenoptic sub-aperture view shuffling for richer color sampling
CN101726829A (en) Method for automatically focusing zoom lens
US9215447B2 (en) Imaging device and imaging apparatus including a pixel with light receiving region on one side of a center of the pixel
CN109164529A (en) The micro- polarization chip arrays and its imaging device of high sampling rate real-time polarization imaging
CN104735349A (en) Synchronous multi-focus Bayer video picture processing system and method
CN104349146A (en) Solid-state imaging device, driving method for the same, and electronic apparatus
CN109934768B (en) Sub-pixel displacement image acquisition method based on registration mode
JP2019521599A5 (en)
US20190379807A1 (en) Image processing device, image processing method, imaging apparatus, and program
CN114913075A (en) Polarization direction full-resolution image reconstruction method

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant