CN102223470A - CCD camera image signal processing method and system - Google Patents

Abstract

The invention provides a new scheme of digital domain filtering for breaking through the bottleneck of noise suppression of the traditional CCD camera, and particularly relates to a new CCD camera image signal processing method and a system. The invention directly digitizes the CCD video signal by using a high-speed high-resolution analog-to-digital converter at a sampling frequency far higher than a reading rate, and then designs a digital signal processing algorithm aiming at the characteristic of reading noise of the CCD video signal to suppress the noise to a sub-electronic level. Currently, large-scale integrated circuit technology, especially high-speed and high-resolution analog-to-digital converter technology and digital signal processing technology, provide a technical foundation for realizing the new idea. Compared with the EMCCD technology, the novel technology has stronger universality, can reduce the CCD reading noise, and provides a technical approach for meeting the requirement of realizing low-speed high-resolution (16bit) by using a high-speed low-resolution analog-to-digital converter (12 bit).

Description

A CCD camera image signal processing method and system

technical field

The invention relates to a CCD camera image signal processing method and system, especially suitable for CCD camera systems with low illumination and weak targets.

Background technique

CCD cameras are widely used, especially in fields that require very demanding image quality, such as medicine, astronomical deep space observation, etc., and have an irreplaceable position. One of the most concerned parameters in the design of CCD cameras is the signal-to-noise ratio (SNR) of the camera. In the case of low illumination and weak targets, readout noise is the main source of noise in CCD cameras. At present, the read noise suppression method is the correlated double sampling method proposed in the 1970s or the improved analog filtering method based on this method, mainly including: correlated double sampling method (CDS), bislope integration method, switching index filtering method, Differential sampling method, reflection delay method, correlation four sampling, etc. These methods have different emphases in use, but they are essentially the same. The principle is to use the slow-changing characteristics of reset noise and adopt related processing techniques, that is, to process the video signal back and forth in one pixel period. Subsamples are then sent to the two input terminals of the differential amplifier to eliminate reset noise while amplifying the video signal. It can be seen that the traditional CCD output noise processing techniques are all carried out in the analog domain. It has many realization forms, but they are all realized by analog circuit, and its frequency domain characteristics are not optimal for CCD signal noise characteristics. The traditional CCD camera video signal processing link block diagram is shown in Figure 1.

The disadvantages of traditional methods are:

1) While suppressing noise, circuit noise is inevitably introduced due to the existence of its own circuit (such as: correlated double sampling circuit).

2) Although integrated chips with correlated double-sampling circuits have been realized, in order to obtain high-performance noise effects, discrete components must be used to build correlated double-sampling circuits, which reduces the reliability of the circuits.

3) As shown in Figure 1, currently the correlated double sampling technology is used to limit the readout noise to 2e-level at low readout rate, and the readout noise is usually 6e-level at 1MHz readout rate, which cannot meet the higher requirements. Application of low illumination and weak target astronomical observation.

Therefore, new ideas and new ways to further reduce readout noise must be considered to improve the dynamic range of optical detectors, especially the ability to detect low-light targets.

Contents of the invention

The purpose of the present invention is to break through the bottleneck of traditional CCD camera noise suppression, and propose a new solution for digital domain filtering, specifically a new CCD camera image signal processing method and system.

Technical idea of the present invention is:

The high-speed and high-resolution analog-to-digital converter is used to directly digitize the CCD video signal at a sampling frequency much higher than the readout rate, and then a digital signal processing algorithm is designed according to the characteristics of the readout noise of the CCD video signal to suppress the noise to the subelectronic level. The current large-scale integrated circuit technology, especially high-speed and high-resolution analog-to-digital converter technology and digital signal processing technology provides a technical basis for the realization of this new idea. Compared with EMCCD technology, this new technology has stronger versatility. It can not only reduce the read noise of CCD, but also provide a way to meet the requirements of low-speed high-resolution (16bit) with high-speed low-resolution analog-to-digital converter (12bit). technical approach.

Technical scheme of the present invention is specifically as follows:

A kind of CCD camera image signal processing method, comprises the following steps:

(1) After the pixel signal output by the CCD is amplified, multi-point sampling is performed on each pixel signal through a high-speed analog/digital conversion circuit; the sampled data is quantized and then output as a digital signal;

(2) Digital signals are latched and synchronized;

(3) Send the synchronously processed digital signal to the acquisition/control computer through the high-speed transmission interface, and the acquisition/control computer obtains the sample signals of all sampling points of each pixel;

(4) Implement filtering algorithm on all sampling points of each pixel in the acquisition/control computer to obtain the pixel value of each pixel, and finally obtain a complete image.

The above-mentioned high-speed transmission interface adopts the Camera link interface protocol.

The filter algorithm described in above-mentioned steps (4) is determined according to the following method: the frame format image that contains the sample signals of all sampling points in each pixel period is converted into a two-dimensional array, and time-domain analysis and spectrum analysis are carried out , so as to determine the appropriate filtering algorithm.

The multi-point sampling described in the above step (1) preferably samples 120-200 sampling points.

The filtering algorithm in the above step (4) is digital domain correlated double sampling, Gaussian weight filtering or various frequency domain filtering algorithms.

A CCD camera image signal processing system provided by the present invention comprises a CCD connected sequentially through a data transmission line, a preamplifier circuit, a high-speed analog/digital conversion circuit, a timing drive and data processing circuit, a high-speed transmission interface and an acquisition/control computer, Among them, from the timing drive and data processing circuit, the high-speed transmission interface to the acquisition/control computer, a control line link is also established.

The above-mentioned high-speed transmission interface is a Cameralink interface circuit.

The above-mentioned high-speed A/D conversion circuit is a 16-bit, 160Msps high-speed A/D converter.

The invention can not only reduce the noise and improve the signal-to-noise ratio of the camera, but also simplifies the hardware design of the camera system and improves the integration degree and reliability of the system. At the same time, with the continuous development of large-scale integrated circuits, it provides a technical basis for the continuous optimization of noise suppression algorithms and the improvement of computing speed. It is specifically manifested in the following advantages:

1. The reliability and integration of the system are improved: the noise filtering scheme adopted in the present invention is carried out in the digital domain, and compared with the traditional design scheme, the relevant double sampling circuit is omitted, which improves the reliability and integration of the system .

2. Flexible design of noise filtering algorithm, strong practicability and versatility: the present invention proposes a new technical solution for direct high-speed digitalization of CCD output video signals. This scheme can design the best digital signal processing algorithm according to the noise characteristics of CCD video signal, so as to obtain better noise suppression ability. The time domain filtering algorithms include: digital domain correlated double sampling, Gaussian weight filtering. In addition, various frequency domain filtering algorithms may also be used.

3. Provide a new concept and general platform for the design of CCD camera system, which greatly reduces the development cost and cycle: new algorithms and new technologies can be implemented based on software, which greatly reduces the development cost and cycle, so that the CCD camera system can keep up with level of technological development. Moreover, the camera system can also modify the filter algorithm and camera control signal in the acquisition software to change the function of the camera system as required.

Description of drawings

Fig. 1 is a typical relationship diagram of the read noise and the read rate of the conventional scheme CCD.

Figure 2 is a block diagram of the traditional CCD camera system video signal processing chain.

Fig. 3 is a structural block diagram of the CCD camera system provided by the present invention.

Fig. 4 is the timing interface circuit of the CCD camera system provided by the present invention.

Fig. 5 is a functional block diagram of the acquisition software of the CCD camera system provided by the present invention.

Fig. 6 is a relationship diagram between readout noise and readout rate of the CCD of the present invention.

Explanation of reference numbers:

1-Pre-amplification circuit, 2-A/D conversion circuit, 3-Sequence drive and data processing FPGA, 4-Camera Link interface, 5-Acquisition/control computer.

Detailed ways

The present invention is a camera system based on a high-speed transmission protocol interface (such as: Camera link interface protocol). The conversion circuit performs sampling. In order to meet the requirements of the later algorithm for sample points, each pixel samples 120 to 200 points, and then through the timing drive and data processing FPGA latch for synchronous processing, and then sent to the acquisition computer through the Camera Link interface. In the acquisition computer, the acquisition software is used to filter the video signal and display the image.

A kind of CCD camera video signal processing method provided by the invention specifically comprises the following steps:

0) After the CCD camera is powered on, under the action of the driving pulse, the CCD outputs the pixel signal;

1) step 0) the video signal output by the CCD is blocked and amplified, so that the output video signal amplitude matches the input signal range of the follow-up analog/digital conversion circuit;

2) Carry out multi-point sampling (generally sampling 120-200 sampling points) to each pixel signal through step 1), output digital signal after quantization;

3) a large amount of data obtained in step 2) is sent to acquisition/control computer by CameraLink interface circuit;

4) Implement a filtering algorithm on the sample signal of each pixel in the acquisition/control computer to obtain the pixel value of each pixel, and display and save the processed image. The filtering algorithm can be digital domain correlated double sampling, Gaussian weight filtering or various frequency domain filtering algorithms.

In order to better illustrate how the present invention is implemented, the present invention will be further described in detail in conjunction with the accompanying drawings and specific embodiments.

A kind of CCD camera system that is used for low illumination, weak target that the present invention provides, referring to Fig. 3, is based on the functional block diagram of CCD camera system provided by the present invention, mainly consists of preamplification circuit 1, analog/digital conversion circuit 2, Sequence driving and data processing circuit 3, high-speed transmission interface circuit (what this program adopts is Camera Link interface, as long as the interface circuit that can satisfy the data transmission rate can be all can), and acquisition control computer 5 is formed, below in conjunction with accompanying drawing 3,4, 5. Introduce the functions of each part in detail:

Pre-amplification circuit 1: the video signal output by the CCD readout amplifier is directly blocked and amplified, so that the amplitude of the output video signal matches the input signal range of the analog/digital conversion circuit 2;

Analog/digital conversion circuit 2: Quantize the pixel signal output by the preamplifier circuit 1, and output digital signals; an analog/digital conversion circuit example is provided in the accompanying drawing 3, and what the present invention adopts is Linglierte Company (Linear Technology)'s analog-to-digital converter LTC2209, LTC2209 is a 16-bit, 160Msps high-performance high-speed analog-to-digital converter with excellent 77.1dB baseband signal-to-noise ratio (SNR) performance and 100dB baseband spurious-free dynamic range (SFDR). It meets the requirements of a large number of samples and high precision.

Timing drive and data processing FPGA circuit 3: generate CCD drive timing signal, digital video signal data output by analog/digital conversion circuit 2 according to a certain timing format requirements and the synchronous control signal of the camera system and send it to the acquisition computer through the Camera link output interface . As shown in accompanying drawing 4, timing drive and data processing FPGA circuit will collect massive data and frame header data (frame header data signal: FHDATA) together in control signal (such as: frame valid signal (FVAL), line valid signal ( LVAL)) sent to the collection control computer in a certain format.

Cameralink interface circuit 4: Through this interface circuit, the collected massive data is transmitted to the acquisition computer at high speed through the acquisition card. And send the control signal sent by the acquisition control computer to the timing drive and data processing circuit 3 to change the working state of the camera (such as adjusting the exposure time, switching the working mode, etc.). An example has been provided in accompanying drawing 3, and the present invention adopts and is the PCIe-1429 acquisition card of NI company, the first image acquisition board based on PCI Express that it National Instruments (National Instruments, be called for short NI) released in January, 2005 The card is suitable for high-throughput vision applications, which can capture images at high speed, high resolution and high pixel depth of Camera Link cameras, and perform complex image acquisition applications.

Acquisition control computer 5: the difference between the acquisition software of the present invention and the traditional acquisition software is that each frame of data collected by the system is no longer an image, but the sample points of all pixel waveforms, and each image must be In order to obtain the real pixel value of the pixel, the filtering process is carried out in the sampling period of the pixel. The acquisition software of this scheme is written by Labview, and the MatLab program is embedded in the acquisition program to facilitate data processing. The functional block diagram of the acquisition software is shown in Figure 5. In the acquisition software, in order to facilitate the analysis of the pixel sampling signal, the obtained frame format image is converted into a two-dimensional array, and time-domain analysis and spectrum analysis are performed on it, which is a filtering algorithm provide a basis for writing.

In addition to conveniently controlling the working state of the camera, the acquisition software also has the following advantages:

1) It is convenient to call the existing display modules and data analysis controls of Labview (such as: waveform display, spectrum analysis, etc.), and the software functions can also be expanded according to needs.

2) Using the powerful matrix operation function of Matlab, the image is quickly restored from the collected sample points through the filtering algorithm written in Matlab. And through the image evaluation module, the filtering algorithm is improved until the user is satisfied.

The following is a further description of the selected Gaussian weight filtering algorithm:

By analyzing the characteristics of the CCD output signal, it is found that within a pixel period of the CCD output signal, the correlation between the reference level and the sampling data near the pixel output is significantly higher than that of the farther sides of it, that is to say, the The sampling point contains more useful information, so assigning a greater weight to the sampling data at this point can effectively suppress the output noise. In this way, the weight given to the sampled data within a pixel period approximates the bell shape of the Gaussian probability distribution function. In order to achieve the effect of digital correlation double sampling, the weight coefficient of the first half of the reference level sampling value is negative, and each The algorithm for each pixel value is shown in formula 1:

$\mathrm{Pix}=\frac{\underset{i=\mathrm{no}}{\overset{2\mathrm{no}-1}{\mathrm{\Σ}}}{\mathrm{\α}}_i{\mathrm{the\; s}}_i}{\underset{i=\mathrm{no}}{\overset{2\mathrm{no}-1}{\mathrm{\Σ}}}{\mathrm{\α}}_i}-\frac{\underset{i=0}{\overset{\mathrm{no}-1}{\mathrm{\Σ}}}{\mathrm{\α}}_i{\mathrm{the\; s}}_i}{\underset{i=0}{\overset{\mathrm{no}-1}{\mathrm{\Σ}}}{\mathrm{\α}}_i}$ (Formula 1)

In the formula, Si is the i-th sampling value, α _i is the weighting coefficient of the i-th sampling point, and α _i =α _(2n+1-i) , 2n sample points are sampled in one pixel period, and the first n One is the number of sampling points of the reference level. From the above operations, it can be seen that directly sampling the pixel data output by the CCD and performing weight filtering can not only obtain the effect of correlated double sampling in the analog circuit, but also save the noise introduced by the correlated double sampling circuit, so that Improve the signal-to-noise ratio of the signal.

Claims (8)

1. CCD camera image signal processing method may further comprise the steps:

(1) the pixel signal of CCD output carries out multi-point sampling by the high speed A circuit to each pixel signal after amplifying; The data that sampling obtains quantize back output digital signal;

(2) digital signal latchs and Synchronous Processing;

(3) send to collection/control computer by the digital signal of high-speed transfer interface after with Synchronous Processing, collection/control computer is obtained the sample signal of all sampled points of each pixel;

(4) all sampled points to each pixel are implemented filtering algorithm in collection/control computer, obtain the pixel value of each pixel, finally obtain a width of cloth complete image.

2. CCD camera image signal processing method according to claim 1 is characterized in that: described high-speed transfer interface adopts Camera link interface protocol.

3. CCD camera image signal processing method according to claim 1 is characterized in that, the described filtering algorithm of step (4) is determined in accordance with the following methods:

The frame format image transitions that will include the sample signal of interior all sampled points of picture dot cycle of each pixel becomes two-dimensional array, carries out time domain analysis and spectrum analysis, thereby determines suitable filtering algorithm.

4. CCD camera image signal processing method according to claim 1 is characterized in that: the described multi-point sampling of step (1) is 120-200 sampled point of sampling.

5. CCD camera image signal processing method according to claim 1 is characterized in that: the described filtering algorithm of step (4) is numeric field correlated-double-sampling, the filtering of Gauss's weights or various frequency domain filtering algorithm.

6. CCD camera image signal processing system, it is characterized in that: comprise CCD, pre-amplification circuit, high speed A circuit, sequential driving and data processing circuit, high-speed transfer interface and the collection/control computer that links successively by data line, wherein, be interfaced to collection/control computer from sequential driving and data processing circuit, high-speed transfer, also establish the control line link.

7. CCD camera image signal processing system according to claim 5 is characterized in that: described high-speed transfer interface is the Cameralink interface circuit.

8. CCD camera image signal processing system according to claim 5 is characterized in that: described high speed A circuit is to have 16, the high speed A device of 160Msps.

Images