CN201726493U

CN201726493U - Deinterlace device specific to surveillance video

Info

Publication number: CN201726493U
Application number: CN2010202470812U
Authority: CN
Inventors: 王伟平
Original assignee: ZHEJINAG PEARMAIN ELECTRONICS CO Ltd
Current assignee: ZHEJINAG PEARMAIN ELECTRONICS CO Ltd
Priority date: 2010-07-02
Filing date: 2010-07-02
Publication date: 2011-01-26
Anticipated expiration: 2020-07-02

Abstract

The utility model relates to the technical field of image processing, in particular to a deinterlace device specific to a surveillance video, comprising an input line caching unit, a storage controller, a storage, an output line caching unit, a motion detection unit and an interpolation unit. The utility model integrates an intra-frame averaging method and a field duplication method, the field duplication method is adopted for the interlacing of a static image, the intra-frame averaging method is adopted for a motion image, motion quantity is calculated specific to the characteristics of a surveillance video, and a processing method is judged by selecting different thresholds according to the type of the surveillance video, therefore not only a picture quality requirement of a monitoring system can be met, but also the processing complexity and power consumption of the deinterlace device can be greatly reduced. The utility model has the advantages that the motion quantity is calculated specific to the characteristics of a surveillance video, and a processing method is judged by selecting different thresholds according to the type of the surveillance video, therefore not only a picture quality requirement of a monitoring system can be met, but also the processing complexity and power consumption of the deinterlace device can be greatly reduced.

Description

A kind of de-interlaced device at monitor video

Technical field

The utility model relates to technical field of image processing, relates in particular to a kind of de-interlaced device at monitor video.

Background technology

At present, used a large amount of simulation interlacing video cameras in the monitoring field, these camera videos as LCD, all need interlacing to handle and just can watch on digital display device during demonstration.

Existing de-interlaced method and apparatus all is to design at the film of DVD or the video in the consumer electronics, and the Video processing that does not have good method and apparatus can satisfy monitoring in supervisory control system requires can well reduce cost and power consumption again.

The commonplace interlace-removing method of current application mainly contains a replica method, the expert method of average of frame, revises along four kinds of algorithms such as adaptive method (ELA), motion compensated filtering algorithms; Wherein: a replica method is exported interleaved odd even two field signal direct splicing with the twice field frequency; This method is not carried out complicated processing to data, realizes easily, and is better to the still image treatment effect, because original things is constant in two relative times, so adjacent two is the strange row and the idol row of things just; But for moving image, because change at two different time things, brightness value is different, and the signal that direct splicing is two will be introduced unacceptable people at object edge and be distortion.The expert method of average of frame is to calculate interpolation with the method that same two row are up and down averaged, and reaches de-interlaced purpose; For the object of motion, when the correlation maximum between the neighbor in the front court, so the expert method of average of frame is because hardware is realized simple, respond well finding broad application, but for the sawtooth (emergence) at static object oblique line edge obviously.The someone had proposed the ELA algorithm again afterwards, and the directional dependency on ELA algorithm utilization edge is gone interlacing to image, and this algorithm can be removed the zigzag edge of image, but was slightly owing not enough aspect the demonstration of picture detail.After ELA, people have proposed the motion compensated filtering algorithm again.This algorithm goes the interlacing effect better, but the algorithm requirement must estimate that it is too high that hardware realizes that cost requires, and only can use accurately to movement locus in a few high-end video equipment.As can be seen, complexity is higher to be caused the very high method of processing cost and power consumption and can't be applied in the supervisory control system from foregoing description, too the method for Jian Huaing and effect can't reach monitoring requirement.Thus, a kind ofly go the proposition of the new method that interlacing is effective, hardware requirement is low necessary.

The utility model content

The utility model is for overcoming the deficiencies in the prior art part, it is a kind of de-interlaced device at monitor video that purpose is to provide method, this device is on the basis of motion detection, combine the expert method of average of frame and a replica method, a replica method is adopted in the interlacing of going to rest image, moving image is adopted the expert method of average of frame, calculate amount of exercise at the monitor video characteristics, and choose different threshold values according to the type of monitor video and come the determination processing method, promptly can satisfy processing complexity and power consumption that the supervisory control system image quality requires can significantly reduce again de-interlacing apparatus, solve problems of the prior art.

The utility model is to achieve the above object by the following technical programs: a kind of de-interlaced device at monitor video comprises the incoming line buffer unit, Memory Controller, memory, output line buffer unit, motion detection unit, interpolating unit; The incoming line buffer unit is used for the line data of the interlaced picture of buffer memory input; Memory Controller is used for after the full delegation of described line buffer unit storage being used to control alternately output at last when the line data of front court and interpolation field by described Memory Controller write memory; Memory is used to store the interlaced image data that described Memory Controller writes; The output line buffer unit is used for reading synchronously the interlaced image data of described memory; Motion detection unit is used to judge which kind of state the interpolation point is in; Interpolating unit is treated interpolation point according to described interpolation point motion state and is carried out the field and duplicate interpolation or frame average interpolation.

The beneficial effects of the utility model: the utility model is on the basis of motion detection, combine the expert method of average of frame and a replica method, a replica method is adopted in the interlacing of going to rest image, moving image is adopted the expert method of average of frame, made full use of the relevant information of vision signal on room and time, level and smooth inadequately, the virtual image in image border when having overcome original device and going interlacing, there are breakpoint, details to show deficiencies such as clear inadequately, not only can have satisfied processing complexity and the power consumption that the supervisory control system image quality required but also can significantly reduce de-interlacing apparatus.

Description of drawings

Fig. 1 is the theory structure block diagram of the utility model de-interlacing apparatus;

Fig. 2 is the de-interlaced flow chart of the utility model;

Fig. 3 is 2 * 3 array of pixels schematic diagrames;

Fig. 4 is the schematic diagram of choosing of interpolation point;

Fig. 5 is monitoring scene and MAD threshold value correspondence table.

Embodiment

Embodiment 1: below in conjunction with specific embodiment the utility model is described further, but protection range of the present utility model is not limited in this:

A kind of de-interlaced device at monitor video as shown in Figure 1, comprises the incoming line buffer unit, Memory Controller, memory, output line buffer unit, motion detection unit, interpolating unit; The incoming line buffer unit is used for the line data of the interlaced picture of buffer memory input; Memory Controller is used for after the full delegation of described line buffer unit storage being used to control alternately output at last when the line data of front court and interpolation field by described Memory Controller write memory; Memory is used to store the interlaced image data that described Memory Controller writes; The output line buffer unit is used for reading synchronously the interlaced image data of described memory; Motion detection unit is used to judge which kind of state the interpolation point is in; Interpolating unit is treated interpolation point according to described interpolation point motion state and is carried out the field and duplicate interpolation or frame average interpolation.

Specifically workflow is as shown in Figure 2: in step S1, from interleaved video data, read adjacent 4 field data, and difference called after c1, c2, c3, c4, wherein c1 is preceding the 3rd field picture, is even field picture; C2 is preceding second field picture, is strange field picture; C3 is preceding first field picture, is even field picture; C4 is current field picture, is strange field picture.Between every adjacent two row of described current strange the c4 of interpolation behavior, so interpolation point is corresponding with described preceding first c3 pixel position.Choosing as shown in Figure 3 of interpolation point, X is the interpolation point.

In step S2, each takes out corresponding data line from described 4 interlaced pictures, as shown in Figure 4, begin to choose 3 neighbor pixels from first pixel of each row of data, counting point from the picture of C2 taking-up is s1, s2, s3, counting point from the picture of C1 taking-up is s4, s5, s6, counting point from the picture of C4 taking-up is s7, s8, s9, counting point from the picture of C3 taking-up is s10, s11, s12, the pixel that takes out among the pixel that takes out among the described c1 and the described c2 is formed one 2 * 3 pixel groups, the pixel that takes out among the pixel that takes out among the described c3 and the described c4 is formed one 2 * 3 pixel groups, as shown in Figure 3.

In step S3, earlier by the formula Y=0.299 * R+0.587 * G+0.114 * B of brightness value, R wherein, it is red that G, B are respectively pixel, green, and blue pixel value obtains the brightness value Y of each pixel of described 2 * 3 array of pixels.The brightness value of described two 2 * 3 each corresponding pixels of pixel groups is asked poor, described each difference is taken absolute value, to described absolute value summation, obtain the MAD value then, formula is:

MAD＝|Y1-Y7|+|Y2-Y8|+|Y3-Y9|+|Y4-Y10|+|Y5-Y11|+|Y6-Y12|

Wherein Y1..Y12 represents the brightness value of s1..s12 pixel in described 2 * 3 array of pixels respectively, is respectively 10,20 as working as Y1..Y12, and 30,40,50,60,70,80,90,100,110,120 o'clock, the MAD value of asking was 360.

In step S4, select different threshold values according to the monitoring scene difference, monitor static scene as fixing, or have only objects of small quantity (people, automobile) when motion at video camera, can select less threshold value (50～100); If voluminous object (people, automobile) motion is arranged, or during rapid movement, then select big threshold value (100～300); When if video camera need move monitoring scene, make scene integral body moving, then select less threshold value (in 50); If the camera sensor noise factor is big (as the CMOS type), then select big threshold value (100～300), threshold value and the described MAD value of choosing compared.

In step S5, here defining the object of which movement coefficient is a, when described MAD value during greater than threshold value, the moving region of described interpolation point for image is described, a=1, get described interpolation in the front court and put the direct addition of pixel value of neighbouring adjacent two pixels and obtain a mean pixel divided by 2 again described, this value will be inserted the pixel value of described interpolation point exactly, and equation expression is:

Xmotion(n，k)＝(X(n-1，k)+X(n+1，k))/2

Wherein X is the pixel value of described interpolation pixel, n is the row at this pixel place in described current field picture, and k is the row at this pixel place in described current field picture, X (n-1, k) and X (n+1 k) is the up and down pixel value of adjacent two pixels in two row of described interpolation point.As (n-1, k)=250, (n+1, k)=200, then the pixel value of described interpolation point is 225 to X as X.

This value is exactly the pixel value that described interpolation point should insert.

When described MAD value during less than threshold value, a=0 illustrates the stagnant zone of described interpolation point for image, inserts the pixel value of the pixel of correspondence among the described preceding first field picture c3 at interpolation point, and equation expression is:

Xstatic(n，k)＝X(n，k)

X (n, k) be the pixel value of pixel among described preceding first c3, wherein X is the pixel value of described interpolation pixel, n is the row at this pixel place in described current field picture, k is the row at this pixel place in described current field picture, (n, k)=300, then the pixel value of described interpolation point is 300 as if X.

Can get thus, the interpolation formula of described interpolation point is expressed as:

Xout(n，k)＝a?Xmotion(n，k)+(1-a)Xstatic(n，k)

As when the MAD value is 360, if threshold value is 300, then the MAD value illustrates that greater than threshold value the interpolation point is in the moving region, this moment a=1, then the pixel value Xout of described interpolation point (n, k)=1*Xmotion (n, k), if Xmotion (n, k) be described 225, then the pixel value Xout of described interpolation point (n, k)=225; If threshold value is, illustrate that the interpolation point is in stagnant zone, at this moment a=0, then the pixel value Xout (n of described interpolation point at 400 o'clock, k)=and 1*Xstatic (n, k), if Xstatic (n, k) be described 300, then the pixel value Xout of described interpolation point (n, k)=300.

In step S6, will be through the synthetic interpolation field picture of described method interpolation with described when the capable alternately output of strange the c4 image in front court, the described line data of working as in the front court of wherein strange behavior, the line data in the described interpolation of the even behavior field.Thus, obtain a frame progressive image.

Above described be specific embodiment of the utility model and the know-why used; if comply with the change that conception of the present utility model is done; when the function that it produced does not exceed spiritual that specification and accompanying drawing contain yet, must belong to protection range of the present utility model.

Claims

1. the de-interlaced device at monitor video is characterized in that: comprise the incoming line buffer unit, Memory Controller, memory, output line buffer unit, motion detection unit, interpolating unit; The incoming line buffer unit is used for the line data of the interlaced picture of buffer memory input; Memory Controller is used for after the full delegation of described line buffer unit storage being used to control alternately output at last when the line data of front court and interpolation field by described Memory Controller write memory; Memory is used to store the interlaced image data that described Memory Controller writes; The output line buffer unit is used for reading synchronously the interlaced image data of described memory; Motion detection unit is used to judge which kind of state the interpolation point is in; Interpolating unit is treated interpolation point according to described interpolation point motion state and is carried out the field and duplicate interpolation or frame average interpolation.