CN102801927A - High-speed image acquiring method - Google Patents

Abstract

The invention discloses a high-speed image acquiring method and relates to the field of high-speed imaging in image detection. The high-speed image acquiring method comprises the steps of: establishing a one-to-one correspondence between each hole and an imaging region corresponding to an image sensor, establishing a global low-speed windowing mode, a local high-speed basic tracking mode and a self-adaptive object tracking mode according to the one-to-one correspondence; respectively acquiring high-seed image sequences by adopting the global low-speed windowing mode and the local high-speed basic tracking mode, or by adopting the global low-speed windowing mode and the self-adaptive object tracking mode according to a third preset condition; and adopting the mode corresponding to the high-speed image sequence with high sampling frequency as a final mode, and using the image sequence obtained through the final mode as a final high-speed image sequence. Through the winding capacity of the image sensor, the image acquisition frequency can be doubled, a certain idle rate is stored, a high signal to noise ratio can be effectively kept, a high-speed image is acquired in real time, the cost and the energy consumption are reduced, and the image distortion is avoided.

Description

A High Speed Image Acquisition Method

technical field

The invention relates to the field of high-speed imaging in image detection, in particular to a high-speed image acquisition method.

Background technique

The high sensitivity of the compound eye is generally defined as a higher temporal resolution, that is, a higher number of sampling frames per second can be obtained. According to the principle of evolution in biology, insects choose the evolution strategy of multi-aperture system to expand the field of view of insects. At the same time, due to the smaller focal length of insect sub-eyes, a smaller image can be obtained, thereby reducing the energy of image processing. consume. In order to obtain higher temporal resolution images, insects can selectively collect images of sub-eyes in different channels at high speed. Compared with global image acquisition, this method can collect images in a targeted manner and increase the number of image sampling frames. , improving the sensitivity of insects.

At present, the realization of a high-speed imaging acquisition system mainly relies on high-speed cameras and multiple low-speed cameras that have been encoded.

In the process of realizing the present invention, the inventor finds that at least the following disadvantages and deficiencies exist in the prior art:

When using a high-speed camera that relies on a high-speed acquisition and cache system, it relies on a high-speed acquisition and cache system, but the cost is high, the energy consumption is large, and the exposure time of each photo is short, so the problem of underexposure is inevitable;

When multiple encoded low-speed cameras are used, it mainly depends on the image acquisition process, the setting of exposure time of different cameras, and the reconstruction of the image in the later stage. Although the cost is low and the image exposure time is sufficient, it is impossible to obtain high-speed images in real time, and a certain degree of distortion will appear after image reconstruction.

Contents of the invention

The present invention provides a high-speed image acquisition method. The present invention realizes real-time acquisition of high-speed images, reduces cost and energy consumption, and avoids image distortion. See the following description for details:

A high-speed image acquisition method, said method comprising the following steps:

(1) Establish a one-to-one correspondence between each sub-eye and the imaging area corresponding to the image sensor, the number of sub-eyes and the number of imaging areas are both N;

(2) Establishing a global low-speed windowing mode through the corresponding relationship;

(3) establish a local high-speed basic tracking mode through the corresponding relationship;

(4) establishing an adaptive object tracking mode through the corresponding relationship;

(5) Using the global low-speed windowing mode and the local high-speed basic tracking mode according to the third preset condition, or using the global low-speed windowing mode and the adaptive object tracking mode to obtain High-speed image sequence: the mode corresponding to the high-speed image sequence with high sampling frequency is used as the final mode, and the image sequence obtained through the final mode is used as the final high-speed image sequence.

The global low speed specifically includes: the image sensor works at a sampling frequency and a sampling interval satisfying a first preset condition.

The establishment of the global low-speed windowing mode through the corresponding relationship specifically includes:

1) Obtain a piece of the image sensor determined by the total number of pixels, the highest sampling frequency and the number of windows;

2) Converting the angular velocity of the moving object into the moving velocity of the imaging area of the object on the image sensor through the corresponding relationship;

3) Obtain the range of the moving speed of the object under the second preset condition;

4) Obtain the number of layers of windowing according to the range of the moving speed;

5) Obtain a windowing mode through the number of layers of windowing, the total number of pixels and the highest sampling frequency;

Wherein, the windowing mode includes: sampling frequency and window position within the window range.

The local high speed specifically means that the size and position of the window change with time, and the sampling frequency is higher than the sampling frequency at the global low speed.

The establishment of a local high-speed basic tracking mode through the corresponding relationship specifically includes:

1) Obtaining the initial imaging position of the object on the image sensor through the global low speed;

2) Determining the basic tracking mode with the current window where the initial imaging position is located as the center;

Wherein, the basic tracking mode is specifically: the sampling frequency of the current window and adjacent windows.

The establishment of an adaptive object tracking mode through the corresponding relationship specifically includes:

1) acquiring the initial imaging position of the object on the image sensor through the global low speed;

2) The windowed area at the next moment is determined by the centroid coordinates of the initial imaging position, and the windowed area is fan-shaped;

张角角平分线与水平轴的夹角θ；3) Determine the radius r and opening angle of the sector

The angle θ between the bisector of the opening angle and the horizontal axis;

4) Determine the size and position of the windowed area.

The beneficial effects of the technical solution provided by the present invention are: compared with the prior art, this method can double the image acquisition frequency through the windowing capability of the image sensor, and at the same time preserve a certain idle rate; it can effectively keep the image at a higher The signal-to-noise ratio is high, and high-speed images are collected in real time, reducing cost and energy consumption; image distortion is avoided through the non-overlapping characteristics of the sub-eye field of view; in addition, the adaptive object tracking mode has a high The learning ability can predict the size and position of the window at the next moment based on the windowed area of the object imaged at the previous moment.

Description of drawings

Fig. 1 is a schematic diagram of global windowing provided by the present invention;

Fig. 2 is the schematic diagram of edge window provided by the present invention;

Fig. 3 is the curve diagram of the relationship between the moving speed of the target and the number of layers of windowing provided by the present invention;

示意图；Fig. 4 shows the three parameters r, θ,

schematic diagram;

估算示意图；Figure 5 is the adaptive windowing area provided by the present invention

Estimation diagram;

FIG. 6 is a flowchart of a high-speed image acquisition method provided by the present invention.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the implementation manner of the present invention will be further described in detail below in conjunction with the accompanying drawings.

Whether in the civilian or military fields, the requirements for the ability to capture high-speed targets are getting higher and higher. At the same time, in special occasions, it is required to achieve high-speed, miniaturized, energy-saving and low-distortion image or video acquisition.

In order to realize real-time acquisition of high-speed images, reduce cost and energy consumption, and avoid image distortion, an embodiment of the present invention provides a high-speed image acquisition method, see Fig. 1, Fig. 2, Fig. 3, Fig. 4, Fig. 5 and Fig. 6 , see the description below:

101: Establish a one-to-one correspondence between each sub-eye and the imaging area corresponding to the image sensor, and the number of sub-eyes and the number of imaging areas are both N;

Among them, before establishing a one-to-one correspondence, it is ensured that the field of view between each sub-eye does not overlap, and there is no blind area. The specific establishment method is well known to those skilled in the art, and will not be described in detail here in the embodiment of the present invention.

102: Establish a global low-speed windowing mode through the corresponding relationship;

Wherein, the global low speed specifically means that the image sensor works at a sampling frequency and a sampling interval satisfying the first preset condition.

Among them, this step specifically includes:

1) Obtain an image sensor with the total number of pixels, the highest sampling frequency and the number of windows determined;

For an image sensor area, assuming that the image sensor is square in shape, the corresponding total number of pixels in the horizontal and vertical directions is M×Mpixel (in this example, M=1024). The upper limit of the sampling frequency of the adopted high-speed data interface is F (the unit is fps). Therefore, the upper limit of the bandwidth is F×M ² pixel.fps (take F=20fps). Suppose the window shape is also square, the size is l×l pixels (take l=256), and the position can be adjusted and changed adaptively. At the same time, k windows can be opened on both horizontal and vertical sides of the image sensor (take k=M/l=4). Therefore, the total number of pixels of the image sensor can be expressed by k ² l ² .

2) Convert the angular velocity of the moving object into the moving velocity of the imaging area of the object on the image sensor through the corresponding relationship;

Among them, after the imaging system is calibrated, the "number of moving windows/second" is uniformly used as the speed v of the object on the image sensor.

3) Obtain the range of the moving speed of the object under the second preset condition;

4) Obtain the number of layers of windowing according to the range of moving speed;

First, assume that the field of view of the fly-eye lens is 180°, corresponding to four windows in the horizontal or vertical direction. First estimate the angular velocity of the object.

When the moving speed of the object is greater than Fk/2 (that is, the number of moving windows/second = 40, and the speed of the corresponding object space is 1800°/s), the global windowing strategy is used to monitor the movement of the object. As shown in Figure 1:

When the moving speed of the object is less than Fk/2 (that is, the number of moving windows/second = 40, and the speed of the corresponding object space is 1800°/s), the edge windowing strategy is adopted at this time, as shown in Figure 2. The number of layers for edge windowing is rounded up to an integer of (k-Fk ² /4v). Here, the angular velocity is taken as 45°/s (that is, the number of moving windows/second = 1), which is slightly greater than the upper limit of the human eye's ability to distinguish angular velocity (35°/s). At this point, take the number of windowing layers as 1.

It can be seen from Figure 3 that in general, the number of windowing layers is one layer, and when the moving speed of the object is relatively high, more than one layer of windowing layers is required. The corresponding relationship between the number of windowing layers and the velocity of the object (including the moving velocity and angular velocity of the object) is shown in Table 1:

Table 1 Comparison table between the number of window layers and the speed of objects (k=4)

The number of edge window layers Object moving speed (number of windows/second) Object angular velocity (°/s) 1 26.7 1202 2 40 1800

As shown in Table 1, when the moving speed of the object is less than 26.7 windows per second, only one layer of edge windows can meet the detection requirements.

5) Obtain the windowing mode through the number of windowing layers, the total number of pixels and the highest sampling frequency.

Wherein, the windowing mode includes: a sampling frequency and a window position within a window range, and the like.

103: Establish a local high-speed basic tracking mode through the corresponding relationship;

Among them, the local high speed is specifically: the size and position of the window change with time, and the sampling frequency is higher than that at the global low speed.

1) Obtain the initial imaging position of the object on the image sensor at a global low speed;

2) With the current window where the initial imaging position is located as the center, determine the basic tracking mode.

Wherein, the basic tracking mode is specifically: the sampling frequency of the current window and adjacent windows.

According to the distance between the adjacent window and the current window, the sampling frequency of the adjacent window is adjusted, so that the closer the distance is, the higher the sampling frequency is, and the farther the distance is, the lower the sampling frequency is, so that the objects flying out of the current window and flying into The direction and sampling frequency of the next window, for a better estimate and balance.

The main thing here is to choose the size and number of windows. Generally speaking, the smaller the window, the more the number of windows, the higher the performance of object tracking, and the more detailed description of details. However, in view of the fact that the image of an object on the image sensor cannot be infinitely small, the number of windows and the cost are generally positively correlated. The size of the window should be larger than the size of the imaging area of the object in the image sensor. Based on this method, it is recommended that the size of the window be 81-100 times the size of the imaging area.

104: Establish an adaptive object tracking mode through the corresponding relationship;

1) Obtain the initial imaging position of the object on the image sensor at a global low speed;

Because the focal length of the fly-eye optical lens is usually small, the magnification of the object is small. When the object is far away from the lens, the imaging area of the target on the detector at each moment in the image can be approximately regarded as a particle.

2) The windowed area at the next moment is determined by the centroid coordinates of the initial imaging position, and the windowed area is fan-shaped;

半径r＞0，张角角平分线与水平轴的夹角θ的绝对角度为0~360°。下一个时刻的加窗区域由之前若干时刻的成像质心的坐标决定，且形状为扇形。该扇形的中心为上一个时刻成像的形心位置O_i(x_i,y_i)。因此，以上三个参量r、θ、可以完备地描述扇形加窗区域的形状。扇形状加窗区域如下图所示：Let the coordinates of the imaging at time i be O _i ( _xi , y _i ). The coordinates O _i+1 (x _i+1 , y _i+1 ) of the object at the next moment fall within the range of a sector, and the opening angle

Radius r>0, the absolute angle of the angle θ between the bisector of the opening angle and the horizontal axis is 0~360°. The windowed area at the next moment is determined by the coordinates of the imaging centroid at several moments before, and its shape is fan-shaped. The center of the sector is the centroid position O _i ( _xi , y _i ) imaged at the previous moment. Therefore, the above three parameters r, θ, The shape of the fanned windowed region can be completely described. The fan-shaped window area is shown in the figure below:

三个参量分别进行分析：In order to analyze the adaptive windowing strategy of fan-shaped window, for r, θ and

Three parameters were analyzed separately:

(1) r is the upper limit value of the moving distance of the object at the next moment relative to the previous moment, which largely determines the size of the windowed area;

(2) Describes the upper limit value of the object's rotation angle at the next moment relative to the previous moment. Together with r, it determines the size of the window. The size of the window is approximately equal to

(3) θ describes the center line of the probability distribution of the area where objects may exist, and the probability of objects appearing on the left and right sides of this line is equal.

The scope and trend of objects appearing in the next moment are not completely determined by the previous moment. In this process, the range at time i+1 is determined by the speed at time i, i-1, and i-2 through fitting, and a certain margin needs to be left (the range needs to give empirical values, that is, according to the speed of the object given by the distribution of the fluctuation range), to ensure that the object will appear in the fan-shaped area at the next moment.

张角角平分线与水平轴的夹角θ；3) Determine the radius r and opening angle of the sector

The angle θ between the bisector of the opening angle and the horizontal axis;

Calculation about θ:

It is generally believed that under the condition of high sampling frequency and the shortest time interval before and after the object, the direction of motion of the object will not change significantly. Therefore, the value of θ here is the angle of the vector formed by the line connecting the centroids at time i and i-1.

About the calculation of r:

The calculation of r is given by the speed of the object at several moments before, and then multiplied by the time interval Δt to be exposed (considering that the number of sampling frames of the object in different windows is different, the speed calculation is used here, not the displacement). The specific method should be based on the size of the speed of the previous few times to complete the fitting and estimation.

的计算：about

The calculation of:

的取值可以较小。但是通常应当留出大约10°的裕度来增加算法的鲁棒性。It is usually estimated by using the prior value of the object's motion direction and the margin required to maintain the robustness of the algorithm. For example, when an object moves along a single-directional trajectory (such as a long straight highway),

The value of can be smaller. But usually a margin of about 10° should be left to increase the robustness of the algorithm.

4) Determine the size and location of the windowed area.

During actual use, the reading area of the image sensor window is usually rectangular. Therefore, the windowed area is the smallest bounding rectangle of the sector.

105: According to the third preset condition, adopt the global low-speed windowing mode and the local high-speed basic tracking mode, or adopt the global low-speed windowing mode and the adaptive object tracking mode to obtain high-speed image sequences respectively; The mode corresponding to the high-speed image sequence is used as the final mode, and the image sequence obtained through the final mode is used as the final high-speed image sequence.

Wherein, the first preset condition, the second preset condition, and the third preset condition are determined according to a situation in an actual application, which is not limited in the embodiment of the present invention during specific implementation.

In summary, the embodiment of the present invention provides a high-speed image acquisition method. This method can double the image acquisition frequency through the windowing capability of the image sensor, and at the same time preserve a certain idle rate; it can effectively keep the image relatively low High signal-to-noise ratio, real-time acquisition of high-speed images, reducing cost and energy consumption; through the non-overlapping characteristics of the sub-eye field of view, image distortion is avoided; in addition, the adaptive object tracking mode has a high It can predict the size and position of the window at the next moment based on the windowed area of the object imaged at the previous moment.

Those skilled in the art can understand that the accompanying drawing is only a schematic diagram of a preferred embodiment, and the serial numbers of the above-mentioned embodiments of the present invention are for description only, and do not represent the advantages and disadvantages of the embodiments.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection of the present invention. within range.

Claims (6)

1. A high-speed image acquisition method, characterized in that the method comprises the following steps: (1) Establish a one-to-one correspondence between each sub-eye and the imaging area corresponding to the image sensor, the number of sub-eyes and the number of imaging areas are both N; (2) Establishing a global low-speed windowing mode through the corresponding relationship; (3) establish a local high-speed basic tracking mode through the corresponding relationship; (4) establishing an adaptive object tracking mode through the corresponding relationship; (5) Using the global low-speed windowing mode and the local high-speed basic tracking mode according to the third preset condition, or using the global low-speed windowing mode and the adaptive object tracking mode to obtain High-speed image sequence: the mode corresponding to the high-speed image sequence with high sampling frequency is used as the final mode, and the image sequence obtained through the final mode is used as the final high-speed image sequence. 2 . The high-speed image acquisition method according to claim 1 , wherein the global low speed specifically includes: the image sensor works at a sampling frequency and a sampling interval that satisfy a first preset condition. 3 . 3. A kind of high-speed image acquisition method according to claim 2, is characterized in that, described through described corresponding relationship, the windowing mode of setting up global low speed specifically comprises: 1) Obtain a piece of the image sensor determined by the total number of pixels, the highest sampling frequency and the number of windows; 2) Converting the angular velocity of the moving object into the moving velocity of the imaging area of the object on the image sensor through the corresponding relationship; 3) Obtain the range of the moving speed of the object under the second preset condition; 4) Obtain the number of layers of windowing according to the range of the moving speed; 5) Obtain a windowing mode through the number of layers of windowing, the total number of pixels and the highest sampling frequency; Wherein, the windowing mode includes: sampling frequency and window position within the window range. 4. A high-speed image acquisition method according to claim 3, characterized in that the local high-speed is specifically: the size and position of the window change with time, and the sampling frequency is higher than the sampling frequency under the global low-speed. 5. A kind of high-speed image acquisition method according to claim 4, is characterized in that, described through described corresponding relationship, establishes the local high-speed basic tracking mode specifically comprises: 1) Obtaining the initial imaging position of the object on the image sensor through the global low speed; 2) Determining the basic tracking mode with the current window where the initial imaging position is located as the center; Wherein, the basic tracking mode is specifically: the sampling frequency of the current window and adjacent windows. 6. A kind of high-speed image collection method according to claim 5, is characterized in that, described through described correspondence, establishes the adaptive object tracking mode to specifically comprise: 1) acquiring the initial imaging position of the object on the image sensor through the global low speed; 2) The windowed area at the next moment is determined by the centroid coordinates of the initial imaging position, and the windowed area is fan-shaped; 3) Determine the radius r and opening angle of the sector

The angle θ between the bisector of the opening angle and the horizontal axis; 4) Determine the size and position of the windowed area.

Images