CN1209919C - Realizing method of digitized processing TV interlaced scanning format conversion - Google Patents

Abstract

The present invention discloses a method for realizing digitized processing TV interlaced scanning format conversion. The line number in each field is increased to 416.5 lines from 312.5 lines via a technique of 4:3 line conversion based on spatial linear interpolation. The field frequency of a static field is increased via a technique of 6:4 field conversion based on frame replication and interpolation. The field frequency of a movable field is increased via a technique of 6:4 field conversion based on average interpolation between fields. A field movement detection technique completes the judgment of whether the current field is movable or static. The precision of the field motion detection is increased via the robustness restriction of moving line numbers. The field frequency of input video is increased to 75Hz via a technique of self-adapting 6:4 field conversion based on detection of field movement. The present invention has the characteristics of simple operation, little resource consumption, good performance, etc. and has the advantages of low performance requirement for television entire circuits, no flicker of show effects and high vertical resolution.

Description

A kind of implementation method of digitized processing TV interlacing scan format conversion

One, technical field

The invention belongs to the VLSI design field, specially refer to the VLSI implementation method of a kind of 50HZ/75HZ interlacing scan format conversion of novel optimization in vision signal scanning and display format conversion and the digital to television.

Two, background technology

Traditional simulated television scan format (Phase Alternation Line system: 50HZ interlacing 625 lines; TSC-system: 60HZ interlacing 525 lines) have two large problems on image display effect: (1) is because field frequency is crossed the low film flicker problem that causes; (2) because the not high problem of picture vertical definition that single game number of scanning lines deficiency is caused.For this reason, when people show in the rear end pixel clock frequency is brought up to 27MHZ by original 13.5MHZ, improve the definition that picture shows by TV signal being carried out multiple digitized processing (interlacing is converted to line by line, frame frequency lifting, figure image intensifying etc.).At present the improved definition television machine that occurs adopts 50HZ line by line or the scan format of 100HZ interlacing mostly.But these improved scan formats can't solve above two problems simultaneously, as the 50HZ progressive format, the field picture of 312.5 row is redeveloped into 625 two field pictures of going, though improved the vertical definition of picture greatly, the frame frequency of 50HZ makes the large-area flicker phenomenon still exist; And the 100HZ interlaced format is promoted to 100HZ with field frequency by original 50HZ, eliminated flicker between the field effectively, but vertical definition is low, and the tangible problem of row structure line still is not resolved.

Three, summary of the invention

According to defective that exists in the above-mentioned background technology and deficiency, the objective of the invention is to, provide a kind of and can effectively eliminate the large-area flicker phenomenon, can improve the implementation method of a kind of digitized processing TV interlacing scan format conversion of vertical definition simultaneously again.

Specific descriptions about 75HZ interlacing scan form are: field frequency is 75HZ, and every main feed line number is 833 row (they being that every field wire number is 416.5 row), interlace mode, and line frequency is about 31.2KHZ.Because the field frequency of 75HZ interlacing scan form is higher than the critical frequency (60HZ) of human eye perceives flicker, single game line number has improved 1/3 than conventional form, thereby taken into account the solution of flicker and vertical definition problem, therefore, under the condition of 27MHZ pixel clock frequency, the 75HZ interlacing is a kind of CRT (cathode-ray tube) scan format of optimum.

The present invention designed a kind of novel optimization towards the hard-wired 50HZ/75HZ interlacing scan of VLSI format conversion method.This method mainly comprises following key technology:

(1) by " based on line translations in 4: 3 of space linear interpolation " technology every field wire number is promoted to 416.5 row by 312.5 row, i.e. 312.5 row * (4/3)=416.5 row;

(2) field frequency of finishing static scene by " duplicating 6: 4 field transformations of interpolation based on frame " technology promotes;

(3) field frequency of finishing moving scene by " based on 6: 4 field transformations of average interpolation between the field " technology promotes;

(4) by " motion detection technique " finish when the pre-treatment field be motion or static judgement, to determine the mode of field transformation next time;

(5) pass through the accuracy that " the robust constraint of motion for line counting " improves a motion detection;

(6) by " based on 6: 4 field transformations of self adaptation of field motion detection " technology the field frequency of input video is promoted to 75HZ, i.e. 50HZ * (6/4)=75HZ.

Described " based on line translations in 4: 3 of space linear interpolation " technology is: carries out the space linear interpolation by per three line data according to 4/3 capable lifting factor and calculates input picture, and the view data of the four lines format of output synchronously and input linear correlation.Can be every of 416.5 row by the every field transformation of 312.5 row so just with the picture format of input video.

Described " duplicating 6: 4 field transformations of interpolation based on frame " technology is: the field frequency that is primarily aimed at static scene promotes.For static scene or change scene slowly in time, the difference between the video sequence consecutive frame is small, can realize that therefore the field frequency under this scene promotes by the way that frame duplicates, and not influence display effect.Specific practice is to carry out the time domain interpolation by per four (being per two frames) images to input video according to 6/4 field frequency lifting factor to handle, the interpolation strategy is per four and duplicates two (being that per two frames duplicate a frame), can obtain six (i.e. three frames) images thus, successively output synchronously.

Described " based on 6: 4 field transformations of average interpolation between the field " technology is: the field frequency that is primarily aimed at moving scene promotes.For having moving object or changing violent scene in time, the difference between the video sequence consecutive frame is huge, is not suitable for adopting aforesaid " duplicating 6: 4 field transformations of interpolation based on frame " technology.Experiment shows, utilizes the temporal correlation between the adjacent parity field, odd even mean field of between interpolation, display effect that can the smooth motion scene.Therefore adopt the way of average interpolation between the field to realize that the field frequency under the moving scene promotes.Specific practice is to handle by per four field picture of input video are carried out the time domain interpolation according to 6/4 field frequency lifting factor, and the interpolation strategy is to insert a mean field in the middle of per two parity field, can obtain six field picture thus, successively output synchronously.

Described " motion detection " technology is: by to pointwise calculating absolute difference between the same sex field of adjacent two frames (Qi Chang and Qi Chang, idol and idol), when absolute difference greater than a certain threshold value, then this is the motor point, otherwise is rest point; When the number in motor point in the delegation during greater than a certain threshold value, then the behavior motion for line, otherwise then be static row; Motion for line in one is counted, when the motion line number greater than a certain threshold value, then this field is a sports ground, otherwise then is static field.Result according to " motion detection " determines the processing mode in field transformation cycle next time.

Described " the robust constraint of motion for line counting " technology is: noise effect and threshold value choice meeting bring deviation to testing result in the motion detection of considering to show up, therefore to when motion for line is counted, adding the robustness constraint, when promptly having only continuous four lines to be motion for line, counter just begins counting.So just can eliminate the influence of isolated motion for line that noise produces to testing result.

Described " based on 6: 4 field transformations of self adaptation of field motion detection " technology is: the field frequency hoisting way that determines to work as the pre-treatment cycle according to the result in last processing cycle " motion detection " adaptively.If detect static scene, then adopt " duplicating 6: 4 field transformations of interpolation based on frame " technology to carry out field frequency and promote; If detect moving scene, then adopt " based on 6: 4 field transformations of average interpolation between the field " technology to carry out field frequency and promote.Like this input video form is transformed to 75 per seconds by 50 per seconds.

The invention provides a kind of implementation method of digitized processing TV interlacing scan format conversion.The hardware implementation structure of this method is simple, and the resource occupation amount is few.By the 75HZ interlacing scan form that obtains after the FPGA platform cumulated volume method, improve scan format with other and compare, have the following advantages:

(1) overcome the large-area flicker phenomenon;

(2) picture exquisiteness, row structure line is not obvious;

(3) static, moving object edge smoother does not have obvious sawtooth;

(4) moving scene is natural and tripping, and static scene is stablized clear.

Four, description of drawings

Fig. 1 is the embodiment of the invention " based on line translations in 4: 3 of a space linear interpolation " principle schematic;

Fig. 2 is the embodiment of the invention " based on line translations in 4: 3 of a space linear interpolation " hardware structure diagram;

Fig. 3 is the embodiment of the invention principle schematic of " duplicating 6: 4 field transformations of interpolation based on frame ";

Fig. 4 is the embodiment of the invention " based on 6: 4 field transformations of average interpolation between a field " principle schematic.

Five, embodiment

Accompanying drawing is specific embodiments of the invention;

Below in conjunction with accompanying drawing content of the present invention is described in further detail:

With reference to shown in Figure 1, be " based on line translations in 4: 3 of space linear interpolation " principle schematic.The video image of input is at first by line translation in 4: 3, thereby realization is by every conversion to every of 416.5 row of 312.5 row.This conversion process can be described by interpolation formula as the formula (1).

$\left\{\begin{array}{c}\mathrm{RA}=\mathrm{ra};\\ \mathrm{RB}=0.25\×\mathrm{ra}+0.75\×\mathrm{rb};\\ \mathrm{RC}=0.5\×\mathrm{rb}+0.5\×\mathrm{rc};\\ \mathrm{RD}=0.75\×\mathrm{rc}+0.25\×{\mathrm{ra}}^{\′};\end{array}\right.---\left(1\right)$

Alphabetical label in the formula (1) is corresponding fully with alphabetical label among Fig. 1.Ra wherein, rb, rc is neighbouring triplex row in the input field picture, ra ' be that next adjacent with rc organized the first trip of triplex row; RA, RB, RC, RD are the new four lines that produces after the conversion, and its spatial distribution as shown in fig. 1.Formula (1) has been determined capable RA, RB, and RC, RD and row ra, rb, rc, the correlation weights between the ra ' and interpolation thereof relation: RA is only relevant with ra, so RA can be duplicated by ra and obtains; RB is relevant with ra and rb respectively, and corresponding correlation weights are respectively 0.25 and 0.75, so RB is that weighted sum by ra and rb obtains; RC is relevant with rb and rc respectively, and the correlation weights all are 0.5, so RC is that weighted sum by rb and rc obtains; RD is relevant with rc and ra ' respectively, and corresponding correlation weights are respectively 0.75 and 0.25, so RD is that weighted sum by rc and ra ' obtains.

Because each the correlation weights in the formula (1) can be expressed as the form of 2 integral number power, so the weighted sum in the formula (1) is calculated and can be realized by shifter-adder.Can simplify the hardware implementation structure of 4: 3 line translation modules like this, improve the operating frequency of system.

Figure 2 shows that the hardware implementation structure figure of " based on line translations in 4: 3 of space linear interpolation " module in the present embodiment.The video data Line in of serial input is the line storage 1 through having FIFO (First-In-First-Out) structure at first.In the present embodiment, dual-port read-write memory DPRAM (Dual-Port RAM) is adopted in the realization of line storage 1, and the degree of depth is 720, and width is 8.Line up is the upstream data passage among Fig. 2, and Line down is a downlink data channel, and the data of adjacent two row of the concurrent respectively transmission of data channel enter 4: 3 capable interpolating circuits 3 up and down, and weighted sum is wherein calculated and realized by dextroposition device 2 and adder 4.In a clock, 3 pairs of inputs of 4: 3 capable interpolating circuits data are calculated according to the relation of the interpolation shown in the formula (1), correspondingly obtain four interpolation results, and in next rising edge clock deposits it respectively in separately d type flip flop 5.Simultaneously, No. four selectors 6 select correct interpolation result to be exported by Line out serial according to Sw sel data select signal from four tunnel input data.Thus, after a field time finishes, obtain the field picture of 416.5 row from output.

Capable interpolation technique in the present embodiment is owing to the spatial coherence of only having considered in the front court in short-term, so there is not motion ghost image problem in the new images that obtains after the conversion; And linear interpolation makes that the display effect of image is better, and it is level and smooth that object edge keeps; Be expert at and adopt the shifter-adder technology in the interpolation, simplified hardware configuration, realize cost thereby reduce.

In the present embodiment, enter the field frequency hoisting module through the data after the line translation.Here adopt " based on 6: 4 field transformations of self adaptation of field motion detection " technology to finish the automatic detection of static or moving scene, and and then realize that the field frequency of input video promotes, and promptly rises to 75HZ by 50HZ.This technology has comprised two kinds of field transformation module and scene detection module at different scenes, promptly " duplicates 6: 4 field transformations of interpolation based on frame " and " based on 6: 4 field transformations of average interpolation between the field ", and " motion detection ".

With reference to shown in Figure 3, at static scene, present embodiment has adopted " duplicating 6: 4 field transformations of interpolation based on frame " technology.The mathematical description of this technology as the formula (2).

(FA，FB，FC，FD，FE，FF)＝

(fa，fb，fc，fb，fc，fd) (2)

Alphabetical label in the formula (2) is corresponding fully with alphabetical label among Fig. 3.Wherein (fc fd) is four sequences (FA, FB, FC, FD, FE, FF) six sequences for exporting of input for fa, fb.Formula (2) has been determined following field interpolation relation: output sequence midfield FA, and FB, FE, FF is a repeat fields, corresponds respectively to a fa, fb, fc, fd; Field FC, FD is the interpolation field, corresponds respectively to a fc, fb.When being input as static scene, frame (fa, fb), frame (fc, fb) and frame (fc is identical fd), so the output image sequence is static effective, but moving scene is had the twitch problem.

With reference to shown in Figure 4, at moving scene, present embodiment has adopted " based on 6: 4 field transformations of average interpolation between the field " technology.The mathematical description of this technology as the formula (3).

(FA，FB，FC，FD，FE，FF)＝

(fa，(fa+fb)/2，fb，fc，(fc+fd)/2，fd) (3)

Alphabetical label in the formula (3) is corresponding fully with alphabetical label among Fig. 4.Wherein (fc fd) is four sequences of input for fa, fb; (FA, FB, FC, FD, FE, FF) six sequences for exporting.Formula (3) has been determined following field interpolation relation: output sequence midfield FA, and FC, FD, FF is a repeat fields, corresponds respectively to a fa, fb, fc, fd; FB, FE is the interpolation field, is respectively average between the field of a fa and fb and a fc and fd.When being input as moving scene, a field fa and a fb, exist movement differential between a field fc and a fd, and interpolation field FB, fa on the scene is distinguished in the effect of FE exactly, between fb and fc, has added the crossing site of crossing on the room and time between fd, make that like this motion of output image sequence is smoother, but there is partial row's jitter phenomenon in static scene.

More than two kinds of field interpolation technology all only be applicable to single scene, and there is defective in another scene.Therefore designed a scene detection module in the present embodiment, i.e. " motion detection ".Image is carried out a motion detection,, field transformation process self adaptation between two kinds of field interpolation technology is switched according to the result who detects." motion detection " by asking for corresponding points in two same sex fields absolute difference and the number in motor point, judge whether this moves; If this is a moving scene, then field interpolation carries out " based on 6: 4 field transformations of average interpolation between the field " next time; If this is a static scene, then field interpolation carries out " duplicating 6: 4 field transformations of interpolation based on frame " next time.

Present embodiment has added " the robust constraint of motion for line counting " in order to guarantee the correctness of scene detection results.Its mathematical description as the formula (4).

Row_motion[n in the formula (4)], n=0,1,2,3 is the capable movement mark of adjacent four lines, wherein represents current line during n=0; Cnt_r_m is motion for line counter (initial value is 0).Formula (4) expression has only when continuous 4 behavior motion for line, and Cnt_r_m just begins to add up, and so just can eliminate the influence to the field motion detection result of isolated motion for line that noise produces.

The mathematical description of " motion detection " as the formula (5).

$\mathrm{Field}\_\mathrm{motion}=\left\{\begin{array}{cc}1;& \mathrm{if}(\mathrm{Cnt}\_r\_m>\mathrm{Threshold}\\ 0;& \mathrm{otherwise}\end{array}\right.---\left(5\right)$

Field_motion is a movement mark in the formula (5); Cnt_r_m is the count value of a middle motion for line; Threshold is the motion for line count threshold.Formula (5) expression: as Cnt_r_m during greater than Threshold, Field_motion be " 1 ", and then working as the front court is sports ground; When Cnt_r_m was less than or equal to Threshold, Field_motion was " 0 ", and then working as the front court is static field.

Then " based on 6: 4 field transformations of self adaptation of field motion detection " technology can be described by formula (6) in the present embodiment.

$\left\{\begin{array}{c}\mathrm{FA}=\mathrm{fa};\\ \mathrm{FB}=(\mathrm{Field}\_\mathrm{motion})?(\mathrm{fa}+\mathrm{fb})/2:\mathrm{fb};\\ \mathrm{FC}=(\mathrm{Field}\_\mathrm{motion})?\mathrm{fb}:\mathrm{fc};\\ \mathrm{FD}=(\mathrm{Field}\_\mathrm{motion})?\mathrm{fc}:\mathrm{fb};\\ \mathrm{FE}=(\mathrm{Field}\_\mathrm{motion})?(\mathrm{fc}+\mathrm{fd})/2:\mathrm{fc};\\ \mathrm{FF}=\mathrm{fd};\end{array}\right.---\left(6\right)$

(fa, fb, fc, fd) four sequences in the formula (6) for importing; (FA, FB, FC, FD, FE, FF) six sequences for exporting; Field motion is a movement mark.Formula (6) expression: output sequence midfield FA, FF is a repeat fields, corresponds respectively to a fa, fd; And (FD FE) for the interpolation field, determines its interpolation content according to the motion of Field as a result of field motion detection for FB, FC.When Field motion was 1, (FD FE) was respectively ((fa+fb)/2, fb, fc, (fc+fd)/2) for FB, FC, and what at this moment adopt is average interpolating method between field at moving scene; When Field motion is 0, (FB, FC, FD, FE) be respectively (fb, fc, fb, fc), what at this moment adopt is that frame at static scene duplicates interpolating method.FA, FB, FC, FD, FE, six of FF be as primary field transformation period, and do a motion detection in output field FF, to determine the mode of field transformation next time.Go round and begin again thus, circular treatment, input video promptly is converted into 75 per seconds.

Finish the conversion that is interlaced to 75HZ interlacing scan form from 50HZ thus, the television image display effect that obtains has flicker free and the high advantage of vertical definition simultaneously, and the hardware of this method realizes that system is not high to the requirement of whole set of television circuit yet.

Claims (1)

1, a kind of implementation method of digitized processing TV interlacing scan format conversion is characterized in that:

1) by " based on the 4:3 line translation of space linear interpolation " technology every field wire number is promoted to 416.5 row by 312.5 row;

2) field frequency of finishing static scene by " duplicating the 6:4 field transformation of interpolation based on frame " technology promotes;

3) field frequency of finishing moving scene by " based on the 6:4 field transformation of average interpolation between the field " technology promotes;

4) by " motion detection " technology finish when the pre-treatment field be motion or static judgement, to determine the mode of field transformation next time;

5) pass through the accuracy that " the robust constraint of motion for line counting " improves a motion detection;

6) by " based on the self adaptation 6:4 field transformation of field motion detection " technology the field frequency of input video sequence is promoted to 75HZ;

Described " based on the 4:3 line translation of space linear interpolation " technology is: carry out the space linear interpolation by per three line data to input picture according to 4/3 capable lifting factor and calculate, obtain and import the view data of the four lines format of linear correlation, output synchronously, can be every of 416.5 row by the every field transformations of 312.5 row with the picture format of input video so just, its transform be suc as formula (1):

Ra in the formula (1), rb, rc is neighbouring triplex row in the input field picture, ra ' is that next adjacent with rc organized the first trip of triplex row; RA, RB, RC, RD are the new four lines that produces after the conversion; 0.25 0.5 and 0.75 etc. is each correlation weights;

Described " duplicating the 6:4 field transformation of interpolation based on frame " technology is: the field frequency at static scene promotes, specific practice is by per four to input video, be per two two field pictures, carrying out the time domain interpolation according to 6/4 field frequency lifting factor handles, the interpolation strategy is per four and duplicates two that promptly per two frames duplicate a frame, can obtain six i.e. three two field pictures thus, output synchronously successively, its transform is suc as formula (2):

(FA，FB，FC，FD，FE，FF)＝

(fa，fb，fc，fb，fc，fd) (2)

(fa, fb, fc, fd) four sequences in the formula (2) for importing; (FA, FB, FC, FD, FE, FF) six sequences for exporting;

Described " based on the 6:4 field transformation of average interpolation between the field " technology is: the field frequency at moving scene promotes, utilize the temporal correlation between the adjacent parity field, odd even mean field of between interpolation, display effect that can the smooth motion scene, specific practice is to handle by per four field picture of input video are carried out the time domain interpolation according to 6/4 field frequency lifting factor, the interpolation strategy is to insert a mean field in the middle of per two parity field, can obtain six field picture thus, output synchronously successively, its transform is suc as formula (3):

(FA，FB，FC，FD，FE，FF)＝

(fa，(fa+fb)/2，fb，fc，(fd+fd)/2，fd) (3)

(fa, fb, fc, fd) four sequences in the formula (3) for importing; (FA, FB, FC, FD, FE, FF) six sequences for exporting;

Described " motion detection " technology is: by the same sex field to adjacent two frames is Qi Chang and Qi Chang, pointwise calculating absolute difference between idol and the idol, when absolute difference greater than a certain threshold value, then this is the motor point, otherwise is rest point; When the number in motor point in the delegation during greater than a certain threshold value, then the behavior motion for line, otherwise then be static row; Motion for line in one is counted, when the motion line number greater than a certain threshold value, then this field is a sports ground, otherwise then is static field, determines the processing mode in field transformation cycle next time according to the result of " motion detection ", its transform is suc as formula (4):

Field_motion is a movement mark in the formula (4); Cnt_r_m is the count value of a middle motion for line; Threshold is the motion for line count threshold;

Described " the robust constraint of motion for line counting " technology is: noise effect and threshold value choice meeting bring deviation to testing result in the motion detection of considering to show up, therefore to when motion for line is counted, adding the robustness constraint, when promptly having only continuous four lines to be motion for line, counter just begins counting, so just can eliminate the influence to testing result of isolated motion for line that noise produces, its transform is suc as formula (5):

Row_motion[n in the formula (5)], n=0,1,2,3 is the capable movement mark of adjacent four lines, wherein represents current line during n=0; Cnt_r_m is the motion for line counter;

Described " based on the self adaptation 6:4 field transformation of field motion detection " technology is: the field frequency hoisting way that determines to work as the pre-treatment cycle according to the result in last processing cycle " motion detection " adaptively, if detect static scene, then adopt " duplicating the 6:4 field transformation of interpolation based on frame " technology to carry out field frequency and promote; If detect moving scene, then adopt " based on the 6:4 field transformation of average interpolation between the field " technology to carry out field frequency and promote, like this input video form is transformed to 75 per seconds by 50 per seconds, its transform is suc as formula (6):

(fa, fb, fc, fd) four sequences in the formula (6) for importing; (FA, FB, FC, FD, FE, FF) six sequences for exporting; Field_motion is a movement mark.

Images