CN101582160A - Foreground detection method and device as well as adaptive threshold adjusting method and device - Google Patents
Foreground detection method and device as well as adaptive threshold adjusting method and device Download PDFInfo
- Publication number
- CN101582160A CN101582160A CNA2009100870831A CN200910087083A CN101582160A CN 101582160 A CN101582160 A CN 101582160A CN A2009100870831 A CNA2009100870831 A CN A2009100870831A CN 200910087083 A CN200910087083 A CN 200910087083A CN 101582160 A CN101582160 A CN 101582160A
- Authority
- CN
- China
- Prior art keywords
- threshold value
- value
- threshold
- slope
- input image
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 83
- 238000000034 method Methods 0.000 title claims abstract description 35
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 15
- 238000012544 monitoring process Methods 0.000 claims abstract description 23
- 238000012360 testing method Methods 0.000 claims description 47
- 238000012545 processing Methods 0.000 claims description 39
- 238000009499 grossing Methods 0.000 claims description 36
- 230000008569 process Effects 0.000 description 13
- 230000008859 change Effects 0.000 description 5
- 230000007774 longterm Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Abstract
The invention discloses a foreground detection method and a device as well as an adaptive threshold adjusting method and a device. The foreground detection method comprises the following steps: comparing a current input image with a background image of each frame to obtain the difference value points respectively corresponding to pixel points of the current input image; carrying out statistics on the percentage of the difference value point number, with value more than different thresholds and choosing a threshold suitable for the current input image according to the variation trend that the percentage of the difference value point increases along with the threshold value. Thus, when foreground detection is carried out, the threshold can be adjusted aiming at the monitoring scene with high or low noise, so that the adjusted threshold can be used for judging the foreground pixel points in the image, thereby improving the accuracy of foreground detection and further improving the accuracy of moving object detection and tracking.
Description
Technical field
The present invention relates to the foreground detection technology, a kind of adaptive threshold control method and the device that particularly can be used for a kind of foreground detection method and the device of moving object detection and tracking and can be used for foreground detection.
Background technology
In the existing video monitoring apparatus, normally utilize static camera to photograph video in the monitoring scene, continuous multiple frames image to this video carries out the moving object detection and tracking then, so that the moving object that is different from background image in the continuous multiple frames image is analyzed.
In the detection and tracking of moving object, foreground detection is first step, and its order of accuarcy directly has influence on the performance of whole device.The processing procedure of existing foreground detection is as follows: earlier current input image and background image are compared, obtain in the current input image the poor of the value of respective pixel in each pixel and background image, the difference that obtains value can be referred to as the difference point, a pixel in the corresponding current input image of each difference point difference; Then the threshold value of each difference point with predefined corresponding specific image noise level compared, and will greater than or more than or equal to difference point pairing pixel in current input image of this threshold value be defined as the foreground pixel point, will smaller or equal to or be defined as the background pixel point less than difference point pairing pixel in current input image of this threshold value.After this, all foreground pixel points are carried out the prospect clustering processing, each foreground area that can obtain being made of different foreground pixel point.
In the practical application, during the environment generation climate change of monitoring scene place, can cause the height of noise in the monitoring scene to change, and should set bigger threshold value, should set less threshold value for the less monitoring scene of noise for the bigger monitoring scene of noise.Yet employed threshold value but all is to preestablish and changeless in the existing foreground detection, can't change with the height of noise in the monitoring scene and dynamically adjust, and this just might reduce the accuracy of foreground detection.
For example, when fine and rainy day or snow day,, may just be not suitable for for rainy day or snow sky at fine preset threshold so in the monitoring scene because the interference in intensity of illumination and the picture has tangible difference; In like manner, use infrared photosensitive sensor at the color sensor of visible light and at night because camera uses by day, thus at daytime preset threshold may just be not suitable for for night.
As seen, at noise or high or low any monitoring scene, existing motion detection is merely able to use changeless threshold value to judge foreground pixel point in the image, thereby can make that the accuracy of foreground detection is not high, and then can make that the accuracy of moving object detection and tracking is not high.
Summary of the invention
In view of this, the invention provides a kind of foreground detection method and device and a kind of adaptive threshold control method and device, can be at the employed threshold value of noise level dynamic adjustments foreground detection in the image.
A kind of foreground detection method provided by the invention comprises:
A1, current input image and background image are compared, obtain and difference point that each pixel of current input image is corresponding respectively;
A2, all difference points are compared with threshold value 1～threshold value n that value increases progressively successively, obtain the number percent of value greater than the difference number of spots of each among threshold value 1～threshold value n respectively, n is the positive integer greater than 1;
A3, in the threshold interval of threshold value i～threshold value j, selected arbitrarily threshold value; Wherein, the value of the number percent that step a2 obtains in the threshold interval of threshold value 1～threshold value i-1 bust and in the threshold interval of threshold value i～threshold value j bust slow down, i is more than or equal to 1 and less than n, j is greater than i and smaller or equal to n;
A4, the selected threshold decision of utilization go out the foreground pixel point in the current input image.
Before described step a1, this method further comprises:
A0, to carrying out walkaway with exist together some frame test patterns of a monitoring scene of described current input image, if detected noise level reaches predetermined degree, then earlier described current input image is carried out picture smooth treatment, then the current input image after the picture smooth treatment is carried out described step a1; Otherwise, directly carry out described step a1.
Described step a0 comprises:
A01, each frame and background image in some frame test patterns are compared, obtain the difference point corresponding respectively with each pixel of each frame test pattern;
A02, each difference point of each frame test pattern is compared with a upper limit threshold and a lower limit threshold value respectively, and statistics obtains in each difference point of each frame test pattern, value is respectively greater than the difference number of spots of upper limit threshold and lower threshold;
A03, calculate each frame test pattern poor greater than the difference number of spots of the difference number of spots of lower threshold and lower threshold, if the mean value of the described difference of each frame test pattern correspondence reaches the predetermined value that an expression noise level reaches predetermined extent, then earlier described current input image is carried out picture smooth treatment, then the current input image after the picture smooth treatment is carried out described step a1; Otherwise, directly carry out described step a1.
Described step a3 comprises:
A31, the value that increases progressively with threshold value 1～threshold value n are that the number percent value that horizontal ordinate, step a2 obtain is an ordinate, the number percent curve that establishment step a2 obtains;
A32, obtain described number percent slope of a curve, and the value that still increases progressively with threshold value 1～threshold value n is horizontal ordinate, makes up slope curve;
A33, in described slope curve, selected and value level off to the threshold interval of the pairing threshold value i of slope～threshold value j of 0, and the slope value of the threshold interval correspondence of threshold value 1～threshold value i-1 is far longer than 0;
The average D of the slope value in the threshold interval that a34, calculating are selected
AvgAnd variances sigma;
A35, according to the resulting average D of step a34
AvgAnd variances sigma, calculate
The result,
Constant for expression Gaussian distribution interval;
A36, in described slope curve, from slope value all threshold values less than step a35 gained result of correspondence, selected value is near of threshold value i.
After the described step a3, before the step a4, this method further comprises: a3 ', be utilized as former frame or the selected threshold value of multiframe, the threshold value that step a3 is selected is carried out smoothing processing;
Described step a4 utilizes the threshold decision after the smoothing processing to go out foreground pixel point in the current input image.
Described step a3 ' is utilized as the selected threshold value of former frame the selected threshold value of step a3 is carried out smoothing processing according to following formula:
T
k’＝(1-β)T
k-1+βT
k；
Wherein, T
k' be the threshold value after the smoothing processing, weight, the T that β sets arbitrarily
K-1Be to be the selected threshold value of former frame, T
kFor the selected threshold value of step a3, k are the positive integer greater than 1.
A kind of foreground detection device provided by the invention comprises:
Difference point acquiring unit is used for current input image and background image are compared, and obtains the difference point corresponding respectively with each pixel of current input image;
The number percent acquiring unit is used for all difference points are compared with threshold value 1～threshold value n that value increases progressively successively, obtains the number percent of value greater than the difference number of spots of each among threshold value 1～threshold value n respectively, and n is the positive integer greater than 1;
Threshold value is selected the unit, is used in the threshold interval of threshold value i～threshold value j, selected arbitrarily threshold value; Wherein, the value of the number percent that the number percent acquiring unit obtains in the threshold interval of threshold value 1～threshold value i-1 bust and in the threshold interval of threshold value i～threshold value j bust slow down, i is more than or equal to 1 and less than n, j is greater than i and smaller or equal to n;
The prospect judging unit is used for the foreground pixel point that utilizes selected threshold decision to go out current input image.
This device further comprises level and smooth decision unit, be used for carrying out walkaway with exist together some frame test patterns of a monitoring scene of described current input image, if detected noise level reaches predetermined degree, then earlier described current input image is carried out picture smooth treatment, and then export the current input image after the picture smooth treatment to described difference point acquiring unit; Otherwise, directly described current input image is exported to described difference point acquiring unit.
Described level and smooth decision unit comprises:
Difference statistics subelement is used for each frame and the background image of some frame test patterns are compared, and obtains the difference point corresponding respectively with each pixel of each frame test pattern;
The quantity statistics subelement, be used for each difference point of each frame test pattern is compared with a upper limit threshold and a lower limit threshold value respectively, and statistics obtains in each difference point of each frame test pattern, and value is respectively greater than the difference number of spots of upper limit threshold and lower threshold;
The enforcement of the judgment subelement, be used to calculate each frame test pattern poor greater than the difference number of spots of the difference number of spots of lower threshold and lower threshold, if the mean value of the described difference of each frame test pattern correspondence reaches the predetermined value that an expression noise level reaches predetermined extent, then earlier described current input image is carried out picture smooth treatment, and then export the current input image after the picture smooth treatment to described difference point acquiring unit; Otherwise, directly described current input image is exported to described difference point acquiring unit.
The selected unit of described threshold value comprises:
Curve is set up subelement, and being used for the value that threshold value 1～threshold value n increases progressively is that the number percent value that horizontal ordinate, number percent acquiring unit obtain is an ordinate, sets up the number percent curve that the number percent acquiring unit obtains;
Slope obtains subelement, is used to obtain described number percent slope of a curve, and the value that still increases progressively with threshold value 1～threshold value n is horizontal ordinate, makes up slope curve;
Interval selected subelement is used at described slope curve, and selected and value level off to the threshold interval of the pairing threshold value i of slope～threshold value j of 0, and the slope value of the threshold interval correspondence of threshold value 1～threshold value i-1 is far longer than 0;
First computation subunit is used to calculate the average D of the slope value in the selected threshold interval
AvgAnd variances sigma;
Second computation subunit, the average D that foundation first computation subunit obtains
AvgAnd variances sigma, calculate
The result,
Constant for expression Gaussian distribution interval;
Relatively choose subelement, be used at described slope curve, from slope value all threshold values less than the second computation subunit gained result of correspondence, selected value is near of threshold value i.
This device further comprises the threshold value smooth unit between selected unit of described threshold value and described prospect judging unit, be used to be utilized as former frame or the selected threshold value of multiframe, and the selected threshold value in the selected unit of threshold value is carried out smoothing processing;
And described prospect judging unit utilizes the threshold decision after the smoothing processing to go out foreground pixel point in the current input image.
Described threshold value smooth unit comprises:
The threshold value storing sub-units is used to be stored as the selected threshold value of former frame;
The level and smooth subelement of carrying out is used for according to following formula, is utilized as the selected threshold value of former frame the selected threshold value in the selected unit of threshold value is carried out smoothing processing:
T
k’＝(1-β)T
k-1+βT
k；
Wherein, T
k' be the threshold value after the smoothing processing, weight, the T that β sets arbitrarily
K-1Be to be the selected threshold value of former frame, T
kFor the selected unit of threshold value is that the selected threshold value of current input image, k are the positive integer greater than 1.
A kind of adaptive threshold control method provided by the invention comprises:
A1, current input image and background image are compared, obtain and difference point that each pixel of current input image is corresponding respectively;
A2, all difference points are compared with threshold value 1～threshold value n that value increases progressively successively, obtain the number percent of value greater than the difference number of spots of each among threshold value 1～threshold value n respectively, n is the positive integer greater than 1;
The threshold interval of a3, selected threshold value i～threshold value j, and in this selected threshold interval, select a threshold value arbitrarily; Wherein, the value of the number percent that step a2 obtains in the threshold interval of threshold value 1～threshold value i-1 bust and in the threshold interval of threshold value i～threshold value j bust slow down, i is more than or equal to 1 and less than n, j is greater than i and smaller or equal to n.
Described step a3 comprises:
A31, the value that increases progressively with threshold value 1～threshold value n are that the number percent value that horizontal ordinate, step a2 obtain is an ordinate, the number percent curve that establishment step a2 obtains;
A32, obtain described number percent slope of a curve, and the value that still increases progressively with threshold value 1～threshold value n is horizontal ordinate, makes up slope curve;
A33, in described slope curve, selected and value level off to the threshold interval of the pairing threshold value i of slope～threshold value j of 0, and the slope value of the threshold interval correspondence of threshold value 1～threshold value i-1 is far longer than 0;
The average D of the slope value in the threshold interval that a34, calculating are selected
AvgAnd variances sigma;
A35, according to the resulting average D of step a34
AvgAnd variances sigma, calculate
The result,
Constant for expression Gaussian distribution interval;
A36, in described slope curve, from slope value all threshold values less than step a35 gained result of correspondence, selected value is near of threshold value i.
After the described step a3, this method further comprises: a3 ', be utilized as former frame or the selected threshold value of multiframe, the threshold value that step a3 is selected is carried out smoothing processing.
Described step a3 ' is utilized as the selected threshold value of former frame the selected threshold value of step a3 is carried out smoothing processing according to following formula:
T
k’＝(1-β)T
k-1+βT
k；
Wherein, T
k' be the threshold value after the smoothing processing, weight, the T that β sets arbitrarily
K-1Be to be the selected threshold value of former frame, T
kFor the selected threshold value of step a3, k are the positive integer greater than 1.
A kind of adaptive thresholding value adjusting device provided by the invention comprises:
Difference point acquiring unit is used for current input image and background image are compared, and obtains the difference point corresponding respectively with each pixel of current input image;
The number percent acquiring unit is used for all difference points are compared with threshold value 1～threshold value n that value increases progressively successively, obtains the number percent of value greater than the difference number of spots of each among threshold value 1～threshold value n respectively, and n is the positive integer greater than 1;
Threshold value is selected the unit, is used for the threshold interval of selected threshold value i～threshold value j, and selectes a threshold value arbitrarily in this selected threshold interval; Wherein, the value of the number percent that the number percent acquiring unit obtains in the threshold interval of threshold value 1～threshold value i-1 bust and in the threshold interval of threshold value i～threshold value j bust slow down, i is more than or equal to 1 and less than n, j is greater than i and smaller or equal to n.
The selected unit of described threshold value comprises:
Curve is set up subelement, and being used for the value that threshold value 1～threshold value n increases progressively is that the number percent value that horizontal ordinate, number percent acquiring unit obtain is an ordinate, sets up the number percent curve that the number percent acquiring unit obtains;
Slope obtains subelement, is used to obtain described number percent slope of a curve, and the value that still increases progressively with threshold value 1～threshold value n is horizontal ordinate, makes up slope curve;
Interval selected subelement is used at described slope curve, and selected and value level off to the threshold interval of the pairing threshold value i of slope～threshold value j of 0, and the slope value of the threshold interval correspondence of threshold value 1～threshold value i-1 is far longer than 0;
First computation subunit is used to calculate the average D of the slope value in the selected threshold interval
AvgAnd variances sigma;
Second computation subunit, the average D that foundation first computation subunit obtains
AvgAnd variances sigma, calculate
The result,
Constant for expression Gaussian distribution interval;
Relatively choose subelement, be used at described slope curve, from slope value all threshold values less than the second computation subunit gained result of correspondence, selected value is near of threshold value i.
This device further comprises:
The threshold value smooth unit is used to be utilized as former frame or the selected threshold value of multiframe, and the selected threshold value in the selected unit of threshold value is carried out smoothing processing.
Described threshold value smooth unit comprises:
The threshold value storing sub-units is used to be stored as the selected threshold value of former frame;
The level and smooth subelement of carrying out is used for according to following formula, is utilized as the selected threshold value of former frame the selected threshold value in the selected unit of threshold value is carried out smoothing processing:
T
k’＝(1-β)T
k-1+βT
k；
Wherein, T
k' be the threshold value after the smoothing processing, weight, the T that β sets arbitrarily
K-1Be to be the selected threshold value of former frame, T
kFor the selected unit of threshold value is that the selected threshold value of current input image, k are the positive integer greater than 1.
As seen from the above technical solution, the present invention can compare each frame current input image and background image, obtain the difference point corresponding respectively with each pixel of current input image, add up the number percent of value then greater than the difference number of spots of different threshold values, and according to the size of difference point number percent along with the variation tendency that the threshold value value increases, choose a threshold value that is suitable for current input image.Like this, when carrying out foreground detection, can regulate threshold value at noise or high or low monitoring scene, thereby can use threshold value after the adjusting to judge foreground pixel point in the image, thereby improve the accuracy of foreground detection, and then can improve the accuracy of moving object detection and tracking.
Further, the present invention also can judge noise intensity overall in the monitoring scene earlier, if detected noise level reaches predetermined degree, noise overall in the expression monitoring scene is stronger, then start the picture smooth treatment that is used for each two field picture of moving object detection and tracking to follow-up, to reduce noise, particularly slight jitter when taking place in video, noise, the minimizing of adopting picture smooth treatment can reduce image are significantly disturbed, thereby further improve the accuracy of foreground detection, and then also can further improve the accuracy of moving object detection and tracking.
Description of drawings
Fig. 1 is the schematic flow sheet of detection noise level in the embodiment of the invention foreground detection method;
Fig. 2 is the schematic flow sheet of threshold value adjustment in the embodiment of the invention foreground detection method;
Fig. 3 is the schematic flow sheet of the selected process of threshold value in the flow process as shown in Figure 2;
Fig. 4 is that threshold value is selected the number percent curve synoptic diagram of setting up in the process as shown in Figure 3;
Fig. 5 is that threshold value is selected the slope curve synoptic diagram of setting up in the process as shown in Figure 3;
Fig. 6 is the structural representation of foreground detection device in the embodiment of the invention;
Fig. 7 is the structural representation in the foreground detection device as shown in Figure 6;
Fig. 8 is the structural representation in the foreground detection device as shown in Figure 6.
Embodiment
For making purpose of the present invention, technical scheme and advantage clearer, below with reference to the accompanying drawing embodiment that develops simultaneously, the present invention is described in more detail.
Being applied to the moving object detection and tracking with foreground detection is example, and in an embodiment, foreground detection method can be mainly two parts:
Pre-service before setting in motion object detection and tracking: the some two field pictures to monitoring scene carry out walkaway, promptly judge overall noise intensity in the monitoring scene earlier, if detected noise level reaches predetermined degree, noise overall in the expression monitoring scene is stronger, then start the picture smooth treatment that is used for each two field picture of moving object detection and tracking to follow-up, to reduce noise, particularly slight jitter when taking place in video, and noise, the minimizing of adopting picture smooth treatment can reduce image are significantly disturbed;
In the threshold value adjustment of setting in motion object detection when following the tracks of: preestablish the different threshold value of a plurality of values, then concerning each frame current input image, current input image and background image are compared, after this, again respectively at the above-mentioned different threshold value of a plurality of values that preestablishes, statistics accounts for the number percent of difference point sum in the current input image greater than the difference number of spots of each threshold value, then according to the size of difference point number percent along with the variation tendency that the threshold value value increases, choose a threshold value that is suitable for current input image and be used for foreground detection.
In above-mentioned two parts, the pretreated preceding part of moving object detection and tracking is optional and nonessential.If foreground detection is applied to other field, need not to carry out the pretreated preceding part of moving object detection and tracking especially.
In addition, need to prove that this paper mentioned background image in full can be long-term background and/or short-term background.Wherein, for long-term background and the short-term background situation of image as a setting, suppose the value B that difference point that A is ordered has corresponding long-term background
A_long(k) and corresponding short-term background another the value B
A_short(k), then the value of definite this difference point is:
Min[|B
A_long(k)-I
A(k) |, | B
A_short(k)-I
A(k) |], I
A(k) be the pixel value of A point in input picture.
Below, the above-mentioned two-part concrete scheme in the present embodiment foreground detection method is elaborated.
1) be used for the preprocessing part of moving object detection and tracking in the foreground detection method of present embodiment:
The interference of many-sided factors such as camera head imaging noise, sharpness, DE Camera Shake, in the input picture of the continuous multiple frames that camera head is gathered, the pixel value of same position background dot can constantly change, if directly this type of input picture is carried out foreground detection, then a lot of noise region mistakes can be identified as foreground area, thereby influence the accuracy of moving object detection and tracking.
But because the variation range and the amplitude of above-mentioned background point are all very little, as long as input picture is carried out smoothing processing, just can reduce the noise of input picture significantly, also eliminate the float in the continuous multiple frames input picture simultaneously, can not cause the appearance of much noise object.But, then can cause negative effect if the less input picture of noise ratio is carried out picture smooth treatment, the easy smoothed background that is treated to of some smaller object for example causes this object can't be detected and follow the tracks of.
Therefore, just need carry out walkaway to some frame test patterns, above-mentioned some frame test patterns and subsequent motion object detection and employed each the frame current input image of tracking belong to same monitoring scene, if detected noise level reaches predetermined degree, then when setting in motion object detection and tracking, startup is to the picture smooth treatment of each frame current input image, so that each frame current input image is carried out foreground detection again through after the picture smooth treatment; Otherwise, when setting in motion object detection and tracking, directly each frame current input image is carried out foreground detection.
More specifically, present embodiment provides a kind of mode of new detection noise level.
Fig. 1 is the schematic flow sheet of detection noise level in the embodiment of the invention foreground detection method.The flow process of detection noise level as shown in Figure 1 at some frame test patterns, and comprises the steps:
Step 101 compares each frame and background image in some frame test patterns, obtains the difference point corresponding respectively with each pixel of each frame test pattern.
Step 102 compares each difference point of each frame test pattern respectively with a upper limit threshold and a lower limit threshold value, and statistics obtains value in each difference point of each frame test pattern respectively greater than the difference number of spots of upper limit threshold and lower threshold.
Need to prove,, comprise noise spot and foreground point usually greater than the difference point of this lower threshold for the less lower threshold of value; And, almost only comprise the foreground point greater than the difference point of this upper limit threshold for the bigger upper limit threshold of value.
For example, with white background and the image that comprises prospect for a width of cloth, wherein except the foreground point, must also comprise owing to take the system noise that camera head the caused point of this image.So, this image and the plain white background image that does not comprise system noise point relatively obtained difference point after, in each the difference point greater than lower threshold, except real foreground point, also comprise the system noise point; And because the gray-scale value of system noise point usually can be less than the foreground point, thereby just only comprise the foreground point greater than possibility in each difference point of upper limit threshold.
Upper limit threshold can be described as prospect threshold value T
Fore, lower threshold can be described as noise threshold T
NoiseCorrespondingly, according to empirical law, in all difference points of each frame test pattern correspondence, greater than noise threshold T
NoiseDifference number of spots N (k, T
Noise) and greater than prospect threshold value T
ForeDifference number of spots N (k, R
Fore) poor, be N (k, T
Noise)-N (k, T
Fore) can regard as noise spot quantity, and the available noise number of spots characterizes the noise level of this frame test pattern.Preferably, prospect threshold value T
ForeDesirable 24, the noise threshold T of empirical value
NoiseEmpirical value desirable 8.
Based on above-mentioned empirical law, step 103, calculate each frame test pattern poor greater than the difference number of spots of the difference number of spots of lower threshold and lower threshold, if the mean value of the described difference of each frame test pattern correspondence reaches a predetermined value, judge that then noise intensity overall in the monitoring scene reaches predetermined degree.
So far, this flow process finishes.
Certainly, the detection noise level can also realize according to existing mode in the present embodiment.
2) the threshold value adjustment part in the foreground detection method of present embodiment:
Set in advance threshold value 1～threshold value n that value increases progressively, i is more than or equal to 1 and less than n, and j is greater than i and smaller or equal to n.
Fig. 2 is the schematic flow sheet of threshold value adjustment in the embodiment of the invention foreground detection method.Threshold value adjustment process as shown in Figure 2, all carry out following steps at each frame current input image:
Step 201 compares current input image and background image, obtains the difference point corresponding respectively with each pixel of current input image.
Current input image in this step can be the current input image after picture smooth treatment, also can be the current input image without picture smooth treatment.
Step 202 compares all difference points of current input image correspondence successively with threshold value 1～threshold value n that value increases progressively, obtain the number percent of value greater than the difference number of spots of each among threshold value 1～threshold value n respectively, and n is the positive integer greater than 1.
The resulting number percent of this step also just equals the ratio of all pixel quantity in difference number of spots and the current input image, the ratio that expression is counted greater than the difference of each among threshold value 1～threshold value n respectively.
The threshold interval of selected threshold value place threshold value i～threshold value j in this step is based on also that foregoing empirical law selectes, and specifies as follows:
The value minimum of threshold value 1, thereby in all difference points of current input image correspondence, value has comprised noise spot and all foreground points all in the current input image basically greater than the difference point of threshold value 1.And, can foreground point quantity then can not change substantially thereupon successively decreasing greater than the noise spot quantity in this and the threshold difference number of spots along with the increasing progressively of threshold value value.Thus, when the value of number percent in the threshold interval of threshold value 1～threshold value i-1 bust and in the threshold interval of threshold value i～threshold value j bust slow down, promptly mean greater than not comprising noise spot substantially in each the difference point among threshold value i～threshold value n, but greater than then might only comprising the part foreground point in each the difference point among threshold value j+1～threshold value n, thereby, just should in the threshold interval of threshold value i～threshold value j, select a threshold value arbitrarily in order neither to comprise noise spot in the difference point that makes greater than selected threshold value, can to comprise all foreground points again.
After this, the threshold decision that can utilize step 203 to select goes out the foreground pixel point in the current input image.
So far, this flow process finishes.
In above-mentioned flow process, the process of the selected threshold value of step 203 can realize based on setting up number percent curve and this number percent slope of a curve curve.As shown in Figure 3, the process of the selected threshold value of step 203 can specifically comprise in the above-mentioned flow process:
What step 2031, the value that increases progressively with threshold value 1～threshold value n were that horizontal ordinate, step 202 obtain is ordinate greater than the number percent value of threshold value 1～threshold value n respectively, sets up the number percent curve.
The number percent curve of setting up in this step can be referring to Fig. 4.In Fig. 4, T represents the value that threshold value 1～threshold value n increases progressively, and supposes that T gets 1～100; R (k, T) expression step 202 obtain respectively greater than number percent value, the 1≤k≤n of threshold value 1～threshold value n.
The slope curve of setting up in this step can be referring to Fig. 5.In Fig. 5, T represents value that threshold value 1～threshold value n increases progressively, supposes that T gets 1～100; D (k, T) expression step 202 obtain respectively greater than number percent value, the 1≤k≤n of threshold value 1～threshold value n.
Step 2034, the average D of the slope value in the threshold interval of threshold value 1～threshold value i-1 that calculating is selected
AvgAnd variances sigma.
Threshold value value selected in Fig. 5 is 21.
So far, this flow process finishes.
In the practical application, can be too inviolent in order to ensure changing at the selected threshold value value of each frame current input image, can after step 203, further be utilized as former frame or the selected threshold value of multiframe input picture in the present embodiment, the threshold value that step 203 is selected is carried out smoothing processing.Concrete threshold value smoothing processing mode can be expressed as formula:
T
k’＝(1-β)T
k-1+βT
k；
Wherein, T
k' be the threshold value after the smoothing processing, weight, the T that β sets arbitrarily
K-1Be to be the selected threshold value of former frame input picture, T
kFor the selected threshold value of step 203, k are the positive integer greater than 1.
Below, again the foreground detection device in the present embodiment is elaborated.
Fig. 6 is the structural representation of foreground detection device in the embodiment of the invention.As shown in Figure 6, the foreground detection device in the present embodiment comprises: level and smooth decision unit 600, difference point acquiring unit 601, number percent acquiring unit 602, the selected unit 603 of threshold value, prospect judging unit 604.
Level and smooth decision unit 600 was used for before the moving object detection and tracking begin, and some frame test patterns are carried out walkaway, and employed each the frame current input image of some frame test patterns and foreground detection belongs to same monitoring scene; If detected noise level reaches predetermined degree, then, earlier current input image is carried out picture smooth treatment, and then export the current input image after the picture smooth treatment to difference point acquiring unit 601 in the setting in motion object detection with after following the tracks of; Otherwise, directly current input image is exported to difference point acquiring unit 601.
Certainly, level and smooth decision unit 600 is for optionally nonessential, and correspondingly, difference point acquiring unit 601 is not can only receive current input image by level and smooth decision unit 600 but can directly receive current input image from camera head just yet.
Difference point acquiring unit 601 is used for current input image and background image from level and smooth decision unit 600 are compared, and obtains the difference point corresponding respectively with each pixel of current input image.
Number percent acquiring unit 602, be used for the difference point of current input image correspondence is compared with threshold value 1～threshold value n that value increases progressively successively, obtain the number percent of value greater than the difference number of spots of each among threshold value 1～threshold value n respectively, n is the positive integer greater than 1.
Threshold value is selected unit 603, is used in the threshold interval of threshold value i～threshold value j, selected arbitrarily threshold value; Wherein, the value of the number percent that number percent acquiring unit 602 obtains in the threshold interval of threshold value 1～threshold value i-1 bust and in the threshold interval of threshold value i～threshold value j bust slow down, i is more than or equal to 1 and less than n, j is greater than i and smaller or equal to n.
Preferably, as shown in Figure 7, level and smooth decision unit 600 can comprise:
Quantity statistics subelement 6002, be used for each difference point of each frame test pattern is compared with a upper limit threshold and a lower limit threshold value respectively, and statistics obtains in each difference point of each frame test pattern, and value is respectively greater than the difference number of spots of upper limit threshold and lower threshold;
Enforcement of the judgment subelement 6003, be used to calculate each frame test pattern poor greater than the difference number of spots of the difference number of spots of lower threshold and lower threshold, if the mean value of the described difference of each frame test pattern correspondence reaches the predetermined value that an expression noise level reaches predetermined extent, then earlier described current input image is carried out picture smooth treatment, and then export the current input image after the picture smooth treatment to described difference point acquiring unit 601; Otherwise, directly described current input image is exported to described difference point acquiring unit 601.
Wherein, the foundation that enforcement of the judgment subelement 6003 is carried out its judgement is as the described empirical law of this paper method part, does not repeat them here.
Preferably, as shown in Figure 8, the processing procedure that the selected unit 603 of threshold value is carried out also can be according to for as the described empirical law of this paper method part and comprise:
Curve is set up subelement 6031, and being used for the value that threshold value 1～threshold value n increases progressively is that the number percent value that horizontal ordinate, number percent acquiring unit obtain is an ordinate, sets up the number percent curve that the number percent acquiring unit obtains;
Slope obtains subelement 6032, is used to obtain described number percent slope of a curve, and the value that still increases progressively with threshold value 1～threshold value n is horizontal ordinate, makes up the slope curve of the value of the slope that obtains;
Interval selected subelement 6033 is used at described slope curve, and selected and value level off to the threshold interval of the pairing threshold value i of slope～threshold value j of 0, and the slope value of the threshold interval correspondence of threshold value 1～threshold value i-1 is far longer than 0;
Relatively choose subelement 6036, be used at described slope curve, from slope value all threshold values less than second computation subunit, 6035 gained results of correspondence, selected value is near of threshold value i.
Still referring to Fig. 6, in the practical application, can be too inviolent in order to ensure changing at the selected threshold value value of each frame current input image, foreground detection device in the present embodiment is between selected unit 603 of threshold value and prospect judging unit 604, also can further comprise threshold value smooth unit 605, be used to be utilized as former frame or the selected threshold value of multiframe, the selected unit of threshold value 603 selected threshold values are carried out smoothing processing.Correspondingly, prospect judging unit 604 utilizes threshold decision after the smoothing processing to go out foreground pixel point in the current input image.
Specifically, threshold value smooth unit 605 can comprise (not shown):
The threshold value storing sub-units is used to be stored as the selected threshold value of former frame;
The level and smooth subelement of carrying out is used for according to following formula, is utilized as the selected threshold value of former frame the selected unit of threshold value 603 selected threshold values are carried out smoothing processing:
T
k’＝(1-β)T
k-1+βT
k；
Wherein, T
k' be the threshold value after the smoothing processing, weight, the T that β sets arbitrarily
K-1Be that the selected unit 603 of threshold value is selected threshold value, the T of former frame
kFor the selected unit 603 of threshold value is that the selected threshold value of present frame, k are the positive integer greater than 1.
Need to prove that difference point acquiring unit 601, number percent acquiring unit 602, the selected unit 603 of threshold value can constitute a threshold value adjustment device, and can be applicable to other purposes except that foreground detection.And this threshold value adjustment device also can further comprise threshold value smooth unit 605.
The above is preferred embodiment of the present invention only, is not to be used to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any modification of being done, be equal to and replace and improvement etc., all should be included within protection scope of the present invention.
Claims (20)
1, a kind of foreground detection method is characterized in that, this method comprises:
A1, current input image and background image are compared, obtain and difference point that each pixel of current input image is corresponding respectively;
A2, all difference points are compared with threshold value 1～threshold value n that value increases progressively successively, obtain the number percent of value greater than the difference number of spots of each among threshold value 1～threshold value n respectively, n is the positive integer greater than 1;
A3, in the threshold interval of threshold value i～threshold value j, selected arbitrarily threshold value; Wherein, the value of the number percent that step a2 obtains in the threshold interval of threshold value 1～threshold value i-1 bust and in the threshold interval of threshold value i～threshold value j bust slow down, i is more than or equal to 1 and less than n, j is greater than i and smaller or equal to n;
A4, the selected threshold decision of utilization go out the foreground pixel point in the current input image.
2, foreground detection method as claimed in claim 1 is characterized in that, before described step a1, this method further comprises:
A0, to carrying out walkaway with exist together some frame test patterns of a monitoring scene of described current input image, if detected noise level reaches predetermined degree, then earlier described current input image is carried out picture smooth treatment, then the current input image after the picture smooth treatment is carried out described step a1; Otherwise, directly carry out described step a1.
3, foreground detection method as claimed in claim 2 is characterized in that, described step a0 comprises:
A01, each frame and background image in some frame test patterns are compared, obtain the difference point corresponding respectively with each pixel of each frame test pattern;
A02, each difference point of each frame test pattern is compared with a upper limit threshold and a lower limit threshold value respectively, and statistics obtains in each difference point of each frame test pattern, value is respectively greater than the difference number of spots of upper limit threshold and lower threshold;
A03, calculate each frame test pattern poor greater than the difference number of spots of the difference number of spots of lower threshold and lower threshold, if the mean value of the described difference of each frame test pattern correspondence reaches the predetermined value that an expression noise level reaches predetermined extent, then earlier described current input image is carried out picture smooth treatment, then the current input image after the picture smooth treatment is carried out described step a1; Otherwise, directly carry out described step a1.
4, as each described foreground detection method in the claim 1 to 3, it is characterized in that described step a3 comprises:
A31, the value that increases progressively with threshold value 1～threshold value n are that the number percent value that horizontal ordinate, step a2 obtain is an ordinate, the number percent curve that establishment step a2 obtains;
A32, obtain described number percent slope of a curve, and the value that still increases progressively with threshold value 1～threshold value n is horizontal ordinate, makes up slope curve;
A33, in described slope curve, selected and value level off to the threshold interval of the pairing threshold value i of slope～threshold value j of 0, and the slope value of the threshold interval correspondence of threshold value 1～threshold value i-1 is far longer than 0;
The average D of the slope value in the threshold interval that a34, calculating are selected
AvgAnd variances sigma; A35, according to the resulting average D of step a34
AvgAnd variances sigma, calculate
The result,
Constant for expression Gaussian distribution interval;
A36, in described slope curve, from slope value all threshold values less than step a35 gained result of correspondence, selected value is near of threshold value i.
5, foreground detection method as claimed in claim 4, it is characterized in that, after the described step a3, before the step a4, this method further comprises: a3 ', be utilized as former frame or the selected threshold value of multiframe, the threshold value that step a3 is selected is carried out smoothing processing;
Described step a4 utilizes the threshold decision after the smoothing processing to go out foreground pixel point in the current input image.
6, foreground detection method as claimed in claim 5 is characterized in that, described step a3 ' is utilized as the selected threshold value of former frame the selected threshold value of step a3 is carried out smoothing processing according to following formula:
T
k’＝(1-β)T
k-1+βT
k；
Wherein, T
k' be the threshold value after the smoothing processing, weight, the T that β sets arbitrarily
K-1Be to be the selected threshold value of former frame, T
kFor the selected threshold value of step a3, k are the positive integer greater than 1.
7, a kind of foreground detection device is characterized in that, this device comprises:
Difference point acquiring unit is used for current input image and background image are compared, and obtains the difference point corresponding respectively with each pixel of current input image;
The number percent acquiring unit is used for all difference points are compared with threshold value 1～threshold value n that value increases progressively successively, obtains the number percent of value greater than the difference number of spots of each among threshold value 1～threshold value n respectively, and n is the positive integer greater than 1;
Threshold value is selected the unit, is used in the threshold interval of threshold value i～threshold value j, selected arbitrarily threshold value; Wherein, the value of the number percent that the number percent acquiring unit obtains in the threshold interval of threshold value 1～threshold value i-1 bust and in the threshold interval of threshold value i～threshold value j bust slow down, i is more than or equal to 1 and less than n, j is greater than i and smaller or equal to n;
The prospect judging unit is used for the foreground pixel point that utilizes selected threshold decision to go out current input image.
8, foreground detection device as claimed in claim 7, it is characterized in that, this device further comprises level and smooth decision unit, be used for carrying out walkaway with exist together some frame test patterns of a monitoring scene of described current input image, if detected noise level reaches predetermined degree, then earlier described current input image is carried out picture smooth treatment, and then export the current input image after the picture smooth treatment to described difference point acquiring unit; Otherwise, directly described current input image is exported to described difference point acquiring unit.
9, foreground detection device as claimed in claim 8 is characterized in that, described level and smooth decision unit comprises:
Difference statistics subelement is used for each frame and the background image of some frame test patterns are compared, and obtains the difference point corresponding respectively with each pixel of each frame test pattern;
The quantity statistics subelement, be used for each difference point of each frame test pattern is compared with a upper limit threshold and a lower limit threshold value respectively, and statistics obtains in each difference point of each frame test pattern, and value is respectively greater than the difference number of spots of upper limit threshold and lower threshold;
The enforcement of the judgment subelement, be used to calculate each frame test pattern poor greater than the difference number of spots of the difference number of spots of lower threshold and lower threshold, if the mean value of the described difference of each frame test pattern correspondence reaches the predetermined value that an expression noise level reaches predetermined extent, then earlier described current input image is carried out picture smooth treatment, and then export the current input image after the picture smooth treatment to described difference point acquiring unit; Otherwise, directly described current input image is exported to described difference point acquiring unit.
As each described foreground detection device in the claim 7 to 9, it is characterized in that 10, the selected unit of described threshold value comprises:
Curve is set up subelement, and being used for the value that threshold value 1～threshold value n increases progressively is that the number percent value that horizontal ordinate, number percent acquiring unit obtain is an ordinate, sets up the number percent curve that the number percent acquiring unit obtains;
Slope obtains subelement, is used to obtain described number percent slope of a curve, and the value that still increases progressively with threshold value 1～threshold value n is horizontal ordinate, makes up slope curve;
Interval selected subelement is used at described slope curve, and selected and value level off to the threshold interval of the pairing threshold value i of slope～threshold value j of 0, and the slope value of the threshold interval correspondence of threshold value 1～threshold value i-1 is far longer than 0;
First computation subunit is used to calculate the average D of the slope value in the selected threshold interval
AvgAnd variances sigma;
Second computation subunit, the average D that foundation first computation subunit obtains
AvgAnd variances sigma, calculate
The result,
Constant for expression Gaussian distribution interval;
Relatively choose subelement, be used at described slope curve, from slope value all threshold values less than the second computation subunit gained result of correspondence, selected value is near of threshold value i.
11, foreground detection device as claimed in claim 10, it is characterized in that, this device is between selected unit of described threshold value and described prospect judging unit, further comprise the threshold value smooth unit, be used to be utilized as former frame or the selected threshold value of multiframe, the selected threshold value in the selected unit of threshold value is carried out smoothing processing;
And described prospect judging unit utilizes the threshold decision after the smoothing processing to go out foreground pixel point in the current input image.
12, foreground detection device as claimed in claim 11 is characterized in that, described threshold value smooth unit comprises:
The threshold value storing sub-units is used to be stored as the selected threshold value of former frame;
The level and smooth subelement of carrying out is used for according to following formula, is utilized as the selected threshold value of former frame the selected threshold value in the selected unit of threshold value is carried out smoothing processing:
T
k’＝(1-β)T
k-1+βT
k；
Wherein, T
k' be the threshold value after the smoothing processing, weight, the T that β sets arbitrarily
K-1Be to be the selected threshold value of former frame, T
kFor the selected unit of threshold value is that the selected threshold value of current input image, k are the positive integer greater than 1.
13, a kind of adaptive threshold control method is characterized in that, this method comprises:
A1, current input image and background image are compared, obtain and difference point that each pixel of current input image is corresponding respectively;
A2, all difference points are compared with threshold value 1～threshold value n that value increases progressively successively, obtain the number percent of value greater than the difference number of spots of each among threshold value 1～threshold value n respectively, n is the positive integer greater than 1;
The threshold interval of a3, selected threshold value i～threshold value j, and in this selected threshold interval, select a threshold value arbitrarily; Wherein, the value of the number percent that step a2 obtains in the threshold interval of threshold value 1～threshold value i-1 bust and in the threshold interval of threshold value i～threshold value j bust slow down, i is more than or equal to 1 and less than n, j is greater than i and smaller or equal to n.
14, adaptive threshold control method as claimed in claim 13 is characterized in that, described step a3 comprises:
A31, the value that increases progressively with threshold value 1～threshold value n are that the number percent value that horizontal ordinate, step a2 obtain is an ordinate, the number percent curve that establishment step a2 obtains;
A32, obtain described number percent slope of a curve, and the value that still increases progressively with threshold value 1～threshold value n is horizontal ordinate, makes up slope curve;
A33, in described slope curve, selected and value level off to the threshold interval of the pairing threshold value i of slope～threshold value j of 0, and the slope value of the threshold interval correspondence of threshold value 1～threshold value i-1 is far longer than 0;
The average D of the slope value in the threshold interval that a34, calculating are selected
AvgAnd variances sigma; A35, according to the resulting average D of step a34
AvgAnd variances sigma, calculate
The result,
Constant for expression Gaussian distribution interval;
A36, in described slope curve, from slope value all threshold values less than step a35 gained result of correspondence, selected value is near of threshold value i.
15, as claim 13 or 14 described adaptive threshold control methods, it is characterized in that, after the described step a3, this method further comprises: a3 ', be utilized as former frame or the selected threshold value of multiframe, the threshold value that step a3 is selected is carried out smoothing processing.
16, adaptive threshold control method as claimed in claim 15 is characterized in that, described step a3 ' is utilized as the selected threshold value of former frame the selected threshold value of step a3 is carried out smoothing processing according to following formula:
T
k’＝(1-β)T
k-1+βT
k；
Wherein, T
k' be the threshold value after the smoothing processing, weight, the T that β sets arbitrarily
K-1Be to be the selected threshold value of former frame, T
kFor the selected threshold value of step a3, k are the positive integer greater than 1.
17, a kind of adaptive thresholding value adjusting device is characterized in that, this device comprises:
Difference point acquiring unit is used for current input image and background image are compared, and obtains the difference point corresponding respectively with each pixel of current input image;
The number percent acquiring unit is used for all difference points are compared with threshold value 1～threshold value n that value increases progressively successively, obtains the number percent of value greater than the difference number of spots of each among threshold value 1～threshold value n respectively, and n is the positive integer greater than 1;
Threshold value is selected the unit, is used for the threshold interval of selected threshold value i～threshold value j, and selectes a threshold value arbitrarily in this selected threshold interval; Wherein, the value of the number percent that the number percent acquiring unit obtains in the threshold interval of threshold value 1～threshold value i-1 bust and in the threshold interval of threshold value i～threshold value j bust slow down, i is more than or equal to 1 and less than n, j is greater than i and smaller or equal to n.
18, adaptive thresholding value adjusting device as claimed in claim 17 is characterized in that, the selected unit of described threshold value comprises:
Curve is set up subelement, and being used for the value that threshold value 1～threshold value n increases progressively is that the number percent value that horizontal ordinate, number percent acquiring unit obtain is an ordinate, sets up the number percent curve that the number percent acquiring unit obtains;
Slope obtains subelement, is used to obtain described number percent slope of a curve, and the value that still increases progressively with threshold value 1～threshold value n is horizontal ordinate, makes up slope curve;
Interval selected subelement is used at described slope curve, and selected and value level off to the threshold interval of the pairing threshold value i of slope～threshold value j of 0, and the slope value of the threshold interval correspondence of threshold value 1～threshold value i-1 is far longer than 0;
First computation subunit is used to calculate the average D of the slope value in the selected threshold interval
AvgAnd variances sigma;
Second computation subunit, the average D that foundation first computation subunit obtains
AvgAnd variances sigma, calculate
The result,
Constant for expression Gaussian distribution interval;
Relatively choose subelement, be used at described slope curve, from slope value all threshold values less than the second computation subunit gained result of correspondence, selected value is near of threshold value i.
19, as claim 17 or 18 described adaptive thresholding value adjusting devices, it is characterized in that this device further comprises:
The threshold value smooth unit is used to be utilized as former frame or the selected threshold value of multiframe, and the selected threshold value in the selected unit of threshold value is carried out smoothing processing.
20, adaptive thresholding value adjusting device as claimed in claim 19 is characterized in that, described threshold value smooth unit comprises:
The threshold value storing sub-units is used to be stored as the selected threshold value of former frame;
The level and smooth subelement of carrying out is used for according to following formula, is utilized as the selected threshold value of former frame the selected threshold value in the selected unit of threshold value is carried out smoothing processing:
T
k’＝(1-β)T
k-1+βT
k；
Wherein, T
k' be the threshold value after the smoothing processing, weight, the T that β sets arbitrarily
K-1Be to be the selected threshold value of former frame, T
kFor the selected unit of threshold value is that the selected threshold value of current input image, k are the positive integer greater than 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910087083 CN101582160B (en) | 2009-06-18 | 2009-06-18 | Foreground detection method and device as well as adaptive threshold adjusting method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910087083 CN101582160B (en) | 2009-06-18 | 2009-06-18 | Foreground detection method and device as well as adaptive threshold adjusting method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101582160A true CN101582160A (en) | 2009-11-18 |
| CN101582160B CN101582160B (en) | 2013-09-18 |
Family
ID=41364296
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200910087083 Active CN101582160B (en) | 2009-06-18 | 2009-06-18 | Foreground detection method and device as well as adaptive threshold adjusting method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101582160B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102222349A (en) * | 2011-07-04 | 2011-10-19 | 江苏大学 | Prospect frame detecting method based on edge model |
| CN103136742A (en) * | 2011-11-28 | 2013-06-05 | 财团法人工业技术研究院 | Foreground detection device and method |
| CN103379268A (en) * | 2012-04-25 | 2013-10-30 | 鸿富锦精密工业（深圳）有限公司 | Power-saving monitoring system and method |
| CN106611417A (en) * | 2015-10-20 | 2017-05-03 | 佳能株式会社 | A method and device for classifying visual elements as a foreground or a background |
| CN109859427A (en) * | 2019-04-04 | 2019-06-07 | 上海天诚比集科技有限公司 | Anti-climbing alarm method based on image variance algorithm |
| CN110233682A (en) * | 2019-06-10 | 2019-09-13 | 同方电子科技有限公司 | A kind of real time threshold signal detecting method |
| CN110555351A (en) * | 2018-06-04 | 2019-12-10 | 瑞昱半导体股份有限公司 | Foreground image extraction method and electronic device |
| US10769799B2 (en) | 2018-08-24 | 2020-09-08 | Ford Global Technologies, Llc | Foreground detection |
| CN113011216A (en) * | 2019-12-19 | 2021-06-22 | 合肥君正科技有限公司 | Multi-classification threshold self-adaptive occlusion detection method |
| US11460851B2 (en) | 2019-05-24 | 2022-10-04 | Ford Global Technologies, Llc | Eccentricity image fusion |
| US11521494B2 (en) | 2019-06-11 | 2022-12-06 | Ford Global Technologies, Llc | Vehicle eccentricity mapping |
| US11662741B2 (en) | 2019-06-28 | 2023-05-30 | Ford Global Technologies, Llc | Vehicle visual odometry |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11783707B2 (en) | 2018-10-09 | 2023-10-10 | Ford Global Technologies, Llc | Vehicle path planning |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1256811A (en) * | 1997-03-17 | 2000-06-14 | 艾利森公司 | Method and apparatus for compensating for click noise in FM receiver |
| CN1415975A (en) * | 2002-10-31 | 2003-05-07 | 复旦大学 | Method for inverse computing digital surficial height from image of synthetic aperture radar in all polarization by one time of flight |
| KR20050060450A (en) * | 2003-12-16 | 2005-06-22 | 삼성전자주식회사 | Overload control apparatus and method using processing load and signalling message's inter-arrival time in a processing board |
| CN101140324A (en) * | 2007-10-11 | 2008-03-12 | 上海交通大学 | Method for extracting sea area synthetic aperture radar image point target |
| CN101266689A (en) * | 2008-04-23 | 2008-09-17 | 北京中星微电子有限公司 | A mobile target detection method and device |
-
2009
- 2009-06-18 CN CN 200910087083 patent/CN101582160B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1256811A (en) * | 1997-03-17 | 2000-06-14 | 艾利森公司 | Method and apparatus for compensating for click noise in FM receiver |
| CN1415975A (en) * | 2002-10-31 | 2003-05-07 | 复旦大学 | Method for inverse computing digital surficial height from image of synthetic aperture radar in all polarization by one time of flight |
| KR20050060450A (en) * | 2003-12-16 | 2005-06-22 | 삼성전자주식회사 | Overload control apparatus and method using processing load and signalling message's inter-arrival time in a processing board |
| CN101140324A (en) * | 2007-10-11 | 2008-03-12 | 上海交通大学 | Method for extracting sea area synthetic aperture radar image point target |
| CN101266689A (en) * | 2008-04-23 | 2008-09-17 | 北京中星微电子有限公司 | A mobile target detection method and device |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102222349A (en) * | 2011-07-04 | 2011-10-19 | 江苏大学 | Prospect frame detecting method based on edge model |
| CN102222349B (en) * | 2011-07-04 | 2013-01-23 | 江苏大学 | Prospect frame detecting method based on edge model |
| CN103136742A (en) * | 2011-11-28 | 2013-06-05 | 财团法人工业技术研究院 | Foreground detection device and method |
| CN103379268A (en) * | 2012-04-25 | 2013-10-30 | 鸿富锦精密工业（深圳）有限公司 | Power-saving monitoring system and method |
| CN106611417B (en) * | 2015-10-20 | 2020-03-31 | 佳能株式会社 | Method and device for classifying visual elements into foreground or background |
| CN106611417A (en) * | 2015-10-20 | 2017-05-03 | 佳能株式会社 | A method and device for classifying visual elements as a foreground or a background |
| CN110555351A (en) * | 2018-06-04 | 2019-12-10 | 瑞昱半导体股份有限公司 | Foreground image extraction method and electronic device |
| CN110555351B (en) * | 2018-06-04 | 2022-02-08 | 瑞昱半导体股份有限公司 | Foreground image extraction method and electronic device |
| US10769799B2 (en) | 2018-08-24 | 2020-09-08 | Ford Global Technologies, Llc | Foreground detection |
| CN109859427A (en) * | 2019-04-04 | 2019-06-07 | 上海天诚比集科技有限公司 | Anti-climbing alarm method based on image variance algorithm |
| US11460851B2 (en) | 2019-05-24 | 2022-10-04 | Ford Global Technologies, Llc | Eccentricity image fusion |
| CN110233682A (en) * | 2019-06-10 | 2019-09-13 | 同方电子科技有限公司 | A kind of real time threshold signal detecting method |
| CN110233682B (en) * | 2019-06-10 | 2022-02-08 | 同方电子科技有限公司 | Real-time threshold signal detection method |
| US11521494B2 (en) | 2019-06-11 | 2022-12-06 | Ford Global Technologies, Llc | Vehicle eccentricity mapping |
| US11662741B2 (en) | 2019-06-28 | 2023-05-30 | Ford Global Technologies, Llc | Vehicle visual odometry |
| CN113011216A (en) * | 2019-12-19 | 2021-06-22 | 合肥君正科技有限公司 | Multi-classification threshold self-adaptive occlusion detection method |
| CN113011216B (en) * | 2019-12-19 | 2024-04-02 | 合肥君正科技有限公司 | Multi-classification threshold self-adaptive shielding detection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101582160B (en) | 2013-09-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101582160B (en) | Foreground detection method and device as well as adaptive threshold adjusting method and device | |
| EP3611915B1 (en) | Method and apparatus for image processing | |
| CN105046655B (en) | A kind of automatic sharpening method of video image and device | |
| CN101582991B (en) | Method and device for processing image | |
| CN106210523B (en) | A kind of exposure adjustment method and device | |
| CN101282425B (en) | Method and device for compensating backlight | |
| CN104184957B (en) | Automatic exposure control method based on adaptive expected image average brightness and suitable for space exploration imaging | |
| CN105635597A (en) | Auto-exposure method and system for vehicle-mounted camera | |
| CN105430267A (en) | Method for adaptively adjusting camera parameters based on face image illumination parameters | |
| CN103327219B (en) | A kind of video image contrast adjustment method and device | |
| CN105516720B (en) | A kind of self-adaptation control method of monitor camera code stream | |
| CN105847708B (en) | Line-scan digital camera automatic exposure method of adjustment based on image histogram analysis and system | |
| CN105100632A (en) | Adjusting method and apparatus for automatic exposure of imaging device, and imaging device | |
| CN108253596B (en) | Air supply method of air conditioner, air conditioner and computer readable storage medium | |
| US8659676B2 (en) | Image analysis device and method thereof | |
| CN103702037B (en) | Automatic regulating method for video image brightness | |
| CN105827993B (en) | Adjust the method and system of image exposure degree | |
| CN104954696A (en) | Automatic EMCCD gain adjusting method | |
| CN110769148A (en) | Camera automatic control method and device based on face recognition | |
| CN106657801A (en) | Video information acquiring method and apparatus | |
| CN105791636A (en) | Video processing system | |
| CN103713652A (en) | Holder rotation speed control method, device and system | |
| CN106546598B (en) | Belt steel surface detection system light source self-adaptation control method | |
| CN106198542A (en) | A kind of WARP-KNITTING technique fast analyser based on smart mobile phone and method | |
| CN110562201A (en) | Control method and control device for windscreen wiper and vehicle |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210127 Address after: No. 602, 6th floor, shining building, 35 Xueyuan Road, Haidian District, Beijing 100083 Patentee after: BEIJING ZHONGXINGTIANSHI TECHNOLOGY Co.,Ltd. Address before: 100083, Haidian District, Xueyuan Road, Beijing No. 35, Nanjing Ning building, 15 Floor Patentee before: Vimicro Corp. |