CN105262947A - Method for improving sampling resolution of photoelectric detector by using multi-channel imaging - Google Patents

Abstract

The invention discloses a method for improving the sampling resolution of a photoelectric detector by using multi-channel imaging. The method uses a compound eye system to perform multi-channel imaging on a target object, and each channel samples the target object at different angles, and makes each channel image exist Sub-pixel offset, and then use the relevant image processing algorithm to fuse the image information collected by each channel to obtain the final image with a spatial sampling density and an effective resolution higher than the sampling resolution of the photodetector. The invention utilizes the imaging characteristics of the compound eye system to improve the sampling resolution of the photodetector without reducing the pixel size of the photodetector.

Description

A Method of Improving the Sampling Resolution of Photodetectors Using Multi-Channel Imaging

technical field

The invention relates to the field of compound eye systems, in particular to a method for obtaining images with high spatial sampling density by using the imaging characteristics of the compound eye system to break through the limitation of the sampling resolution of photodetectors.

Background technique

The photodetector is an important part of the optical imaging system. In the case of certain optical system parameters, the size of its target surface determines the imaging field of view, and the size of its pixel is an important factor affecting the resolution of the imaging system. The sampling of the target object by the photodetector satisfies the Nyquist sampling theorem.

In a traditional vision system, the spatial resolution of the entire imaging system is limited by the sampling resolution of the photodetector (that is, the pixel size) and the resolution of the optical system. At the level of existing optical design, the resolution of the optical system is often not the main factor limiting the resolution of the entire imaging system. Therefore, improving the sampling resolution of photodetectors has become an important technical approach to improve the spatial resolution of imaging systems. Traditionally, improving the sampling resolution of a photodetector means reducing the pixel size of the photodetector. Reducing the pixel size of the photodetector can suppress the aliasing blur caused by the undersampling of the photodetector, and further tap the potential of the optical system. However, under the existing material and technology level, the photodetector pixel size has almost reached the limit (several microns), if the photodetector pixel size is still reduced under the current technology level, the shot noise will be greatly increased . In addition, reducing the pixel size of the detector and increasing the density of detection pixels can improve the image resolution, but at the same time, it will reduce the amount of photons obtained by each pixel, thereby reducing the image signal-to-noise ratio and resulting in a decrease in image quality. The cell size cannot be reduced infinitely.

In order to solve the above problems, the imaging characteristics of the compound eye system can be used to obtain multi-channel images of the observation target, and the image information collected by each channel can be fused by using the relevant image processing method according to the sub-pixel offset of each channel image, and the obtained A final image with a spatial sampling density and effective resolution higher than the photodetector sampling resolution. Therefore, the spatial sampling density and effective resolution of the photodetector can be improved without reducing the pixel size, breaking through the limitation that photodetector sampling must satisfy the Nyquist sampling theorem.

The method involved in the invention overcomes the contradiction between the photodetector pixel size and sampling resolution, and uses the imaging characteristics of the compound eye system and related image processing methods to improve the photoelectric The sampling resolution of the detector.

Contents of the invention

The technical problem to be solved by the invention is: in the traditional vision system, improving the sampling resolution of the photodetector (reducing the pixel size) is an important technical way to improve the spatial resolution of the imaging system. However, under the current material and technology level, the photodetector pixel size has almost reached the limit (several microns), and reducing the photodetector pixel size and increasing the detection pixel density can improve image resolution, but At the same time, it will reduce the amount of photons obtained by each pixel, thereby reducing the image signal-to-noise ratio and resulting in a decrease in image quality, so the pixel size cannot be reduced infinitely. Using the imaging characteristics of the compound eye system and related image processing methods can improve the sampling resolution of the photodetector without reducing the pixel size.

The technical solution adopted by the present invention to solve the technical problem is: a method for improving the sampling resolution of the photodetector by using multi-channel imaging, using the compound eye system to perform multi-channel imaging on the target object, and according to the sub-pixel offset of each channel image The correlation image processing method is used to fuse the image information collected by each channel to obtain the final image whose spatial sampling density and effective resolution are higher than the sampling resolution of the photodetector. This method does not reduce the pixel size of the photodetector. Under the condition of , improving the sampling resolution of the photodetector includes the following steps:

(1) Use a pre-designed compound eye system to image the target object to obtain a multi-channel image, and at the same time make each channel image have a sub-pixel offset;

(2) According to the sub-pixel offset between the images of each channel, a correlation image processing method is used for fusion processing to obtain a final image with higher spatial sampling density and effective resolution than the images of each channel.

Wherein, the compound eye system in the step (1) includes a flat compound eye system and a curved compound eye system.

Wherein, the pre-designed compound eye system in the step (1) means that the compound eye system used must meet specific imaging requirements, that is, each channel image of the compound eye system has a sub-pixel offset for the same target.

Wherein, the step (2) sub-pixel offset exists in each channel image of the compound eye system means that the difference between the position deviation of the images of the same target object in each channel relative to the center of the image is a non-integer multiple of the pixel size , which can be less than one cell size, or an integer multiple of the cell size plus less than one cell size.

Wherein, in the step (2), the differences in the positional deviations of the images of the same object formed by each channel between two pairs of the images relative to the center of the image may be equal or unequal.

Wherein, in the step (1), according to the sub-pixel offsets between the images of each channel, the image processing method is used to perform fusion processing, which means that the image pixels of each channel are processed according to the imaging offsets of each channel to the same target. Rearranging and merging into one image breaks through the limitation of the sampling resolution of the photodetector used in the compound eye system, and obtains a final image with a higher spatial sampling density and effective resolution than the images of each channel.

The principle of the present invention is: a method for improving the sampling resolution of photodetectors by using multi-channel imaging, using the imaging characteristics of the compound eye system to obtain multi-channel images of the observation target, and using the correlation method according to the sub-pixel offset of each channel image In the image processing method, the image information collected by each channel is fused to obtain a final image with a spatial sampling density and an effective resolution higher than the photodetector sampling resolution.

The present invention has the following advantages: the method involved in the present invention utilizes the imaging characteristics of the compound eye system and related image processing methods to improve the sampling resolution of the photodetector, and can increase the space of the photodetector without reducing the pixel size. Sampling density and effective resolution break through the limitation of Nyquist sampling theorem for photodetector sampling.

Description of drawings

Fig. 1 is the basic flowchart of the present invention.

Fig. 2A and Fig. 2B are the ideal target function, the image of the sine function with period P=0.01mm and its detailed enlarged view. Figure 2A shows a sine function with period P=0.01mm (image space frequency 100lp/mm), and the coordinate range is (-0.15mm～0.15mm), and Figure 2B shows its partial expansion in the range of (-15μm～15μm) .

Figure 3A to Figure 3J are the sampling of the ideal objective function by 5 channels and the corresponding detailed enlarged images, Figure 3A is the image obtained by sampling the ideal objective function in channel 1, and the coordinate range is (-0.15mm～0.15mm), Figure 3B is Its local expansion diagram in the range of (-15μm～15μm); Figure 3C is the image obtained by sampling the ideal target function in channel 2, the coordinate range is (-0.15mm～0.15mm), and Figure 3D shows it in (-15μm～15μm) The partial expansion diagram of the range; Figure 3E is the image obtained by sampling the ideal target function in channel 3, the coordinate range is (-0.15mm～0.15mm), and Figure 3F is its partial expansion diagram in the range of (-15μm～15μm); Figure 3G It is the image obtained by sampling the ideal target function in channel 4, and the coordinate range is (-0.15mm～0.15mm). Figure 3H is its partial expansion in the range of (-15μm～15μm); Figure 3I is the sampling of the ideal target function in channel 5. The obtained image has a coordinate range of (-0.15 mm to 0.15 mm), and Fig. 3J is a partial expanded view of it in the range of (-15 μm to 15 μm).

Figure 4A and Figure 4B are the reconstructed image obtained by image processing based on 5 channel sub-images and its detailed enlarged view, Figure 4A is the reconstructed image obtained by image processing based on 5 channel sub-images, the coordinate range is (-0.15 mm～0.15mm); Figure 4B shows its partial expansion in the range of (-15μm～15μm).

detailed description

In order to make the purpose of the present invention, technical scheme and advantage clearer, in conjunction with the method involved in the present invention, according to optical system and photoelectric detector parameter, have carried out simulation analysis to adopting multi-channel image to improve detector sampling resolution, with this The present invention is further described in detail.

The following is the simulation analysis process and results of using multi-channel images to improve the sampling resolution of the detector.

(1) Compound eye imaging system simulation

The compound eye system includes an array lens and a photodetector array. The array lens diameter is 4mm×4mm square, the focal length is 50mm, the center distance of the array lens is 4mm×4mm, the number of photodetector pixels is 4872×3248, and the pixel size is 7.4μm×7.4 μm, so the size of the photosensitive surface is 36.05mm×24.04mm, and the Nyquist sampling frequency of the detector is 67.56lp/mm.

In the simulation experiment system, the object distance is 3000mm, the image distance is 50.847mm according to the Gaussian formula, and the image height is 4mm (corresponding to the object space 236mm).

(2) Objective function simulation

In order to facilitate the analysis without loss of generality, in the simulation experiment, the target object function is a one-dimensional sine function distribution, and through the ideal optical system imaging, the target image function on the image plane is also a one-dimensional sine function distribution.

Figure 2A shows a sine function with period P=0.01mm (image space frequency 100lp/mm), and the coordinate range is (-0.15mm～0.15mm), and Figure 2B shows its partial expansion in the range of (-15μm～15μm) .

(3) Channel image acquisition

In the simulation imaging experiment, a compound eye system with 5 sub-channels is selected. The photodetector can simultaneously collect 5 sub-images of the simulated target on the image surface of the array lens. The amount of displacement, so the horizontal range of each sub-image is (-0.15mm~0.15mm). In addition, in order to facilitate the display of image details, the partial expansion diagram of each sub-image in the range of (-15 μm to 15 μm) is also given correspondingly.

3A to 3J are the sub-images of the target collected by each sub-channel and the corresponding detailed enlarged views. Since the Nyquist sampling frequency of the photodetector is 67.56lp/mm, for the target object, since its spatial frequency is 100lp/mm, which is greater than the Nyquist sampling frequency of 67.56lp/mm of the photodetector, the acquired Undersampling of images, i.e. sub-image sampling is limited by the Nyquist sampling frequency of the detector.

(4) Image processing algorithm

It can be seen from FIG. 3A to FIG. 3J that each simulated sub-image has a sub-pixel offset. From the perspective of information acquisition, because each sub-image has a sub-pixel displacement, that is, it contains complementary information, and a suitable image registration processing algorithm can be used to reconstruct a high spatial frequency target image that exceeds the limit of the Nyquist sampling frequency of the detector. Figure 4A and Figure 4B are the reconstructed images obtained through processing, which are in good agreement with the target image.

Simulation experiments show that using the multi-channel sub-images obtained by the compound eye imaging system, using appropriate image processing algorithms can reconstruct high-spatial-frequency target images that exceed the Nyquist sampling frequency limit of the detector, and improve the ability of the image to distinguish details. The sampling resolution of the photodetector is improved without reducing the pixel size of the photodetector.

The above is only a specific implementation mode in the present invention, but the scope of protection of the present invention is not limited thereto. Anyone familiar with the technology can understand the conceivable transformation or replacement within the technical scope disclosed in the present invention. All should be covered within the scope of protection of the claims of the present invention.

Claims (6)

1. A method for improving the sampling resolution of photodetectors by using multi-channel imaging, using a compound eye system to perform multi-channel imaging on the target object, and adopting a correlation image processing method according to the sub-pixel offset of each channel image to collect each channel into The image information is fused to obtain the final image whose spatial sampling density and effective resolution are higher than the sampling resolution of the photodetector. This method improves the sampling resolution of the photodetector without reducing the pixel size of the photodetector. , characterized in that it includes the following steps: Step (1), using the pre-designed compound eye system to image the target object to obtain a multi-channel image, and at the same time make each channel image have a sub-pixel offset; Step (2), according to the sub-pixel offsets between the images of each channel, a correlation image processing method is used to perform fusion processing to obtain a final image with higher spatial sampling density and effective resolution than the images of each channel. 2. A method for improving the sampling resolution of a photodetector by using multi-channel imaging according to claim 1, wherein the compound eye system in step (1) includes a plane compound eye system and a curved surface compound eye system. 3. A method for improving photodetector sampling resolution by multi-channel imaging according to claim 1, characterized in that: the pre-designed compound eye system in step (1) means that the compound eye system used must meet certain requirements. The imaging requirements of the compound eye system, that is, there is a sub-pixel shift in the image of each channel of the compound eye system for the same target. 4. A method for improving photodetector sampling resolution by using multi-channel imaging according to claim 3, characterized in that: wherein, each channel image of the compound eye system has a sub-pixel offset which means that each channel pairs the same The difference between the position deviation of the target image relative to the center of the image is a non-integer multiple of the pixel size, which can be less than one pixel size, or an integer multiple of the pixel size plus less than one pixel size. 5. a kind of method utilizing multi-channel imaging to improve photodetector sampling resolution according to claim 1, is characterized in that: each channel described in the step (2) is opposite to the image formed by the same target object in pairs The differences in the position deviation amounts of the centers may be equal or unequal. 6. A kind of method utilizing multi-channel imaging to improve photodetector sampling resolution according to claim 1, is characterized in that: described in step (1) adopts image according to the sub-pixel offset between each channel image Processing method Fusion processing refers to rearranging and merging the image pixels of each channel into one image according to the imaging offset of each channel to the same target, so as to break through the limitation of the sampling resolution of the photodetector used in the compound eye system and obtain spatial sampling A final image with a density and effective resolution higher than the photodetector sampling resolution.