GB2274371A

GB2274371A - Measurement and control using motion classification flags

Info

Publication number: GB2274371A
Application number: GB9324678A
Authority: GB
Inventors: Kenneth Stanley Jones
Original assignee: Individual
Current assignee: Individual
Priority date: 1992-12-02
Filing date: 1993-12-01
Publication date: 1994-07-20
Anticipated expiration: 2013-12-01
Also published as: GB9324678D0; GB2274371B

Abstract

Flags indicating the type of motion at each pixel of an image, as determined by comparing successive interframe differences are used to determine the size and speed of movement of an object within the image and to generate other control signals. <IMAGE>

Description

DIGITAL IMPULSE RESPONSE r FILTER Te SYSTEItI- This technical description is for digital electronic video measurement metrics and control systems, and follows from the method of the finite temporal domain impulse response filter for communications and control as given of prior applications. These are "Bandwidth reduction employing a DATV channel" GB 2265783 (digital impulse response modulation), and "Video image motion classification and display" GB 2265784 (the video digital impulse response filter) which uses as appropriate ('Y', 'U' or 'V') luminances for digital TV/video system techniques. These are analogous with the CCIR 601 video system standard and similar systems.

The purposes and uses provide digital metric values for physical measurements, control communications, value reading display, and deviation indications. There are flag pulse rates and boundary transition edges for relative velocity speed measurand values derived from object dynamics in the kinematic image. These video kinematic controls may be for remote functions at a distance without the need of a special object contact environment, or physical instrumental intrusion.

Summary of Digital Video Criteria for the TV input signal.

The video source picture signal inputs of Figure 2 item blocks 1 & 2 are assumed to be conventionally frame-scanned, x > along, and Y

down, using consecutive pixel cell (of Y, U, V) sampling of the objective scene along a fast horizontal scan line, then more quickly back and down for the start of the next pixel scan line, all in frame and scan order. n temporal interframe temporal period At, between successive linearly corresponding frame pixel cells, sequentially provides two intrafield pixel absolute difference iS P or PAD-values for p,q. Intrafield mean difference relationships of the inlet 3-frame cascade are of Figures la/b, and Figure 2 are of the discrete filter item blocks 1/2/3/4 with 5 producing binary p,q values.

The overall video 4-channel communication pulse codification CbIltoll ] of Figure 2/9 is for signalling, measurement and control as implementable in fast semiconductor logic technique harduear, incorporating submicron ( < r) silicon or' similar logic gate function architectures for very large-scale integrated manufacturing methods.

The (Y,U/V) Luminance-based DIR-Filter System input (review).

The Figure la/b cross-section illustrations of digital vision signal values are of (n-l)bit/pixel image motion detection amplitudes of PAD-value (or A#P variable) from item blocks 1 to 3. The computation which is given by a 2's complement difference is of prior definition (GB 2265783/4).

The primary (n-l)bit input method uses an interframe interval At input of pixel values, item blocks 1 & 2, as separate digital inputs with the block 2 output signal at first or prime dilation delay across a t, for item block 3, to complement the single intrafield mean Pixel Absolute Difference-value. The (n-l)bit aP signal is given by change detection within the interframe a t period. The resultant video signal form is the finite frame intrafield or luminance density difference. The (n-l)bit per pixel intrafield representation of motion or change detection feeds into item blocks 4a/b for the Absolute Difference Limit Threshold -transfer in item block 4a.The binary ADLT-transfer as a bit rate reduction transform is output as a 1-bit discrete variable '0'/'1', and follows a comparison with the ADLT-value (v) outcome of item block 4b as the fixed or variable (v) reference value. The comparison ADLT-value (v) may follow signal amplitudes and/or external threshold reference values.

This initial q output can feed into a Median type 3.SP.

1-bit mode delay filter over three oP. pixel cell periods of Figure 8. The 1-bit output 'q' feeds into a serial 1-bit/pixel temporal transfer 'doublet' ot. as delay line item block 5. This contains the discrete signal tensor element change Eq., x' y 'p , x, E [ 'p',x,y ] / The digital pixel frame signal criterion for each video image motion detection condition via a 'q' > 'P' hamiltonian dilation is for an inlet of three consecutive frames imported as Pk(n+1):Pk(n):Pk(n-1) , etc.

At the temporal 'doublet' end point ot. the image digits represent temporal image dilation as pixel tensor element E'p',x, y ] in parallel synchronism back to ['q',x'',y''3; the following interframe value of the (n-l)bit 1#DLT-transfer comparison on output from the st. 'doublet' delay line. Hence there are simultaneous output 1-bits of 'q' & 'p',with 'p' having been held to hamiltonian dilation. These distinct motion change detection variables of 'p' and 'q' are analytic from pixel representations within a 3-frame or 3-field/frame period of the 'doublet'. Each imported 3-frame cascade Pk(n) pixel identity provides one input bP condition into the 4-way virtual flag b(m) channel filter.The channel pulse processes organise the both or each change, the two individual dual differential 'aut' change, and the null union 2-input boolean AND relationships as AND (1), -/NOT AND (2), NOT/- AND (3) with NOT/NOT AND (4) gate function combinations arranged within the architectural structure of item blocks 6/9. This virtual method is for the characteristic singleton signal transition-edge of a flag output (eg O - 5 Volts f The quaternary defined channel pulse is uniquely of one b(m) motion class admitted from set [ bi1to4# compared against the ADLT-value efficiency characteristic |h|for for effective detection in rejecting noise and interference.

The visually observed screen image movement conditions discerned from a viewed picture display, as seen on a television set, monitor or screen, of Figure 3 are as given in this electronic image system method.

Only an illustrative description of embodiment is given for these digital function signal b(m) flag devices. Figure 2 item blocks 6/9 correspond to the resultant range [ bi1to4#. n positive boolean logic function in convention and notation is assumed for either positive ( I ) or negative ( t ) transition edges, equivalent embodiments (...) are not discussed.

system Control Method - Illustrations By way of concept illustration of the materials the following set of drawings are used for reference purposes: Figure la - intrafield PAD-values quantified into binary differences predestined as p and q in cross-section of the video frame scan of the digital impulse response filter (DIRF) prior to ADLT-transfer. The lower axis edge shows frame Line number (140 - 240), while Luminance axis (0 - 250) is shown as ordinate (Y).

Figure lb - The specific image segmentation by motion class CbIlto4 ] is for the imported 3-frame cascade on output from the temporal 'doublet' as 'q' R 'p'.

Figure 2 - Main input circuit processes and connection morphism showing the item block numbered stages for the video digital impulse response filter (DIRF) together with the linear and speed measurement, and control device system.

Figure 3 - Visual display screen image movement properties and characteristics relating to video object motion class b(m).

Figure 4 - Basic boolean gate architecture and (up-)counter arrangements for (DIRF) relative speed or walking 'gait' (velocity) measurement, for DlRS/G.

Figure 5 - Basic boolean gate arrangements for video DIRF aiming of a timing trigger pulse for DIRQ targetting.

Figure 6 - Basic boolean gate and function arrangements for video DIRF image object 1clearance' transition edge DIRC.

Figure 7 - The (5-)input boolean (N)AND gate function 'onto' heuristic logic for item block 10 details, to control the measurand clock for DIRM - linear flags EbI6to11#.

Figure 8 - The Median type qE 'q'change mode ancillary filter.

Figure 9 - The fast DIRF 'gait' measurement upon velocity by DIRM - speed onto the (5-)input boolean (N)RND gate function.

The ADLT-transfer inequality to 1-bitXpixel is prior to the input of the 'doublet' temporal ot. delay line and significantly determines q in overall system sensitivity by the ADLT-value (v), for reliability, noise and interference free instrumentation measurements. This is an evaluation according to ambient environmental lighting and source object or other luminousity ('Y' - signal related) values which are effective upon the camera and/or vision inlet signal. The exact concept parametericisation requirements needed for DIRF systems accordingly matches the necessary features of the product functional design requirement and anticipated ambient conditions such as actual luminousity values for the Luminance (Y) amplitudes expected of the Pkfn (-/+) 1) 'doublet'.

The system methods for device of this digital impulse response filter (DIRF) system are as follows: The DIRF speed or walking 'Gait' velocity measurements (DIRS/G), Figures 1/4 illustrate.

The pixel or pel cell flag pulse frequency of the Figure 2 item block 7 output is also as item block 7 in Figure 4 for the b(2) pulse rate during the 'uncovering' image condition within the video signal. The pulses are resultant from the boolean filter achitecture for b(2) by the temporal logical (p n q) q) > TRUE t 2 + '1' equation using the differential logic -/NOT AND gate function arrangement of item block 7 [ c.f. the forward feed digital impulse response m-descriptor for employment of a DATV channel3.

The horizontal object speed of the image boundary edge is represented by the actual quantity of b(2) 'uncovering' pixel cell item 18 flags fed into item block 10 during only one line scan period, the line rate. For a given digital video/TV system parameters the faster the relative image object speed during the line scan period # t the greater the number (or quantity of b(2/3) flag pulses counted. With a higher performance line scan speed the productivity of singleton flag pulses b(2/3) is higher which leads to a greater precision in the metric count as item 28 for display.

The pixel cell rate singleton b(2) for the 'uncovering' image condition results distinctly in 1-bit metrics, as item 26 in Figure 4 (DIRG). These are fed into the bistable (up-)counter item block 15 to increment the output binary count of motive pixel cells given as the item 27 metric. The enabling 3-input boolean AND gate function process is activated for a single pre-settable line scan by number (N, from N1#N2) N2) through item block 24 with also the manual disable switch as item block 25. The pixel flag count total for a specified line scan as item 27 is reached at the end of a b(2) flag count period.

The object image speed as given by the accumulated value item 27 is in arbitary bitstream (walking 'gait' movement or speed) units from counter item block 15.

The single frame line value N may be changed to cover a small range of line scans Ns N2 for a total integrated value over a number of scan lines [ N2 - N11 N2 > N## for when the image object edge speed remains constant over the choosen vertical range (N1 e N2).

The absolute count values achieved from item block 15 also depend upon the scene object image projection relationship with the camera scanning system method well as the behaviour of the image object under observation. The instrument calibration relationship for the b(2) flag bitstream count is based upon relative velocity 'rationalised' by item 16 an optional binary transcoder RAN, as the conversion look-up table, or mapping correspondence. This is incorporated prior to the BCD converter and display chip items block 17 for flag pulse b(5). items block 16 a 17 may be integral. This gives a continuing steady flag count reading for the walking 'gait' or object speed from the incrementation (up-)counter item block 15. item block 29 is to provide the auto/manual disable/reset function switch for the digital display.

The scene object image 'covering' condition b(3) as item 19 class flags from item block 8 of Figure 2 may be used in full substitution for the b(2) flags of item 18 to produce the speed counter metric, p q) 3 v '1' at flag presence for pulse b(5) totalisation.

DIRF Aiming (DIRA), Figures 1/2/3 & 5.

The target aiming filter, Figures 1/2/3 & 5, show the aim channel image condition b(6) flag as item 30. The flag b(6) is produced by a 2-input AND gate function of item block 34 from the b(3) image condition item 19, the flag b(3) from the initial scene object 'covering' conditions. When spatially ( ~ p n q)v 4 + '1' the digital filter detection of pulse b(4) as item 31 is gated from object 'stationary' conditions into item block 9 the flag pulse is held in the spatial interpixel registerE4#, as item block 11. This registerC4 ] store is for holding pulse item 33 to enable the 2-input boolean AND gate function item block 34 upon advancement of a scene object. The flag b(3) pulse item 19 flags give the fast asynchronous transition-edge flag b(6) as output item 30.

Stationary image motion conditions on p R q exist immediately prior to establishing the 'covering' leading-edge boundary detection of the flag b(6) transition as a trigger firing pulse, item 30. The transition timing of the flag b(6) pulse item 30 for whatever purpose exists within the video frame of reference for the display picture Pk(n) of the cascade.

The DIRF period measurement, Figures 1/3 & 6, by the DIRC spatio-temporal sT4 counting vector.

The hamiltonian variable dilation formation (p A q) #1 = > '1' through item block 5 is for the 'fully motive' image condition into item block 6 produces b(1) flag pulses as item 42 to feed into the 3-input boolean AND gate function. When the gate item block 10 is also enabled by pulses from item block 24; by the pre-settable line number or numbers (N or Ni EN2) together with an auto/manual break-enable switch item block 36 of Figure 6 into the 3-input boolean AND gate function item block 10.The signal resultant item 20 starts the following asynchronous bistable (up-)counter item block 15 which accumulates a line-period metric count value item 27 for the spatio-temporal vector gTo. The vector gTr value resolves during a video line scan period of the scene object or body.

The period time resulting as item 27 has metric units in transfer flow into item block 16, the binary transcoding ROM for standard calibration conversion. The measurand path flow is then to item block 17 for BCD conversion and digital units display. Again item blocks 16 and 17 may be an integral unit.

The method is in similarity to that already detailed through to Figure 4, and may include the auto/manual reset item 29.

AdditionallY the vector sampling period for sT is extendable over a range or batch of line numbers (N1 (N1#Np#) N2 ) for an intergral count value.

The video DIRC measurement proportionality is with the linear horizontal object dimension (say length) as scanned during time period sTr. This quantity amounts to a number of b(1) 'fully motive' pixel cell image condition pulses resultant during a single or batch numbered line scan within a video frame. This flag measurand bitstream is metric to the true object dimension or length of the video image object.

The DIRF for Linear Measurement, Trigger Control, and Remote indication (DIRM - Linear), Figures 1 to 6 & 7.

The use of all four asynchronous channel pulse outputs EbI1to43 as Figure 1/2 & 7 from item blocks 6 to 9 is necessary of the precision 8T1,# counting vector for the period measurement and pulse control filter. An additional high-frequency clock as pulse item 39 from bistable item block 14 is optionally used to enhance the resolution value of the measurand sTt,+ resultant metric, as signal item 20. Sn external visual indication or light (bY a LED or otherwise) of object direction or orientation is with respect to the horizontal line scan normal x > is given according to pulse flag b(2) item 18 of 'uncovering' for leaving, and If for flag b(3) of 'covering' advancement.

The foundation method principles incorporated into the [ bi7to1## flag devices have already been suggested here in outline within the earlier description. This is in evidence were a common method, block nomenclature and configuration detail exists from Figures 4 through to 6, and onto 2 & 7.

To instigate the transversal sTi,a + time period measurement a boolean 5-input (N)AND gate function basis is defined as item block 10, and controlled by the high frequency clock item 39s from item block 14, to produce the penultimate measurand item 20 bitstream from item block 10 in Figures 2 & 7. The bitstream incrementated metric resultant produced as item 20 is bistable (up-)counted by item block 15 for an accumulation or total value as item 21 of Figure 2 for display. The signal input to the calibration ROM converter item block 16 is resultant in the 'rational' value units as item 22. The 'rational' value is fed to the BCD converter and onto the digital display as item block 17 for the final readout values upon the accumulation value item 20.

Mode of Measurand Operation.

The flag b(4) 'fully motive' class pulse as item 31 may be fed into the interpixel spatial registerE4 ] as item block 11 to provide the held 'start' (DIRT) function item 33 between pixel cells, and the failure indicator feed item 32.

Simultaneously the 2-input boolean OR gate function, as item block 35, is fed from either a b(2) flag of the 'uncovering' class pulse item 18, the image condition from item block 7, or a b(3) flag of the 'covering' class pulse item 19, the image motion condition from item block 8 for either to result in flag b(7). This summed 'aut' enable flag follows either flag b(2) or b(3) and is for representing a single intrafield change in image conditions as item 36 for output indication of a motion change state. This pulse is embedding appropriately the logic of items 18 or 19. The single intrafield 'aut' flag b(7) as item 36 is pulsed into a boolean 2-input AND gate function together with the b(4) held item 33 pulse from the spatial pixel intercell registerE4 ] as item block 11.The register [ 4 ] output pulse is fed into the 2-input boolean AND gate function item block 42. The output pulse as item 30 is the flag class b(8). This firing trigger pulse b(8) has been 'aut' embedded from flags b(2) or b(3). The composite flag b(8) pulse is the 'start' pulse for the 5-input (N)AND gate function of item block 10 for measurement, distant signalling, etc.

The later composite 'stop' pulse flag b(9) of item 40 is normally kept referenced 'high' or 'on' for the (N)AND gate function item block 10 as an enablement upon the measurand pulse bitstream of item 20.

In sensing an image motion object scan transversal #T1,# period either the b(2) pulse or the b(3) pulse upon a 'aut' function is fed into item block 35. The 2-input boolean OR gate function as item block 35 results in a b(7) 'aut' flag as item block 36. This is fed to the upper 2-input boolean NAND gate function as item block 38 to provide the resultant 'stop' pulse transition into the 5-input (N)QND gate function of item block 10. The second input to this 5-input (N)AND gate function of item block 38 is the interpixel cell register (1 ] of item block 34 when resulant in a spatial hold as item 37 of the earlier b(1) pulse for 'fully motive video image motion conditions at item 41.This produces as resultant the 'stop' pulse b(9) upon the image edge transition b(1) as item 40.

The flag b(9) also provides for the pulse edge transition, and indication of image object clearance during an interframe period (DIRC).

The 5-input (N)AND gate function as item 10 is de-enabled to switch off the 8T1,4 measurand bitstream at the boundary edge condition. The 4th and 5th input to the (N)AND gate functions are the line count number (N) pulse item as block 24 which is an AND gate function with the auto/manual de-enable switch item block 36 for bitstream flow establishment.

The high frequency clock EC ] item block 14 may be at the digital TV/video frame pixel/pel bit-rate frequency as item 35 to provide a higher magnitude to measurement resolution. Most simply however flag b(1) pulses from item block 9 as item 31 provide low resolution pulses item 39 for item block 10 input.

median type 'q' -*'q' change mode ancillary filter.

The crucial pulse quality of q 1-bits, from Figure 2 item blocks 4 a/b, (or any b(m) flag pulse) is optionally enhanced by inserting the median type filter, as illustrated in Figure 8, into the digital pulse system circuit. The input pulse q quantum is fed into the interpixel spatial cell delay of SP.

of item block 1 for a resultant output 'q. When and if 'q is followed by a consecutive q flag pulse the consequent penultimate OR gate function is productive of the 'q' resultant output, the substantiated flag pulse item 8 (or pulse b'(m) for the Majority filtration position following transmission).

The single occurrence of one q 1-bit pulse failure within the input signal bitstream quota of four is a 1 in 4 error, in any order of succession. The median filter will maintain a restituted output 'q' pulse as output item 8 with a more certain consistency. The error correction reinforcement function is provided by the tree of three 2-input AND gate functions as items 2, 4 & 6 with each spanning across the three spatial interpixel cell (3. & .) delays. Each ahd every AND gate spatial function throughput is fed into the one 3-input OR gate function. With three of four correct q (or b(m)) based flag pulses present in the distributed 3.bop.

delay line, at a, b, c or d, an output 'q' pulse is resultant as item 8. The hierarchial outlet OR gate function signal is fed back into the temporal 'doublet' transfer delay-line illustrated earlier in Figure 2 as item block 5.

The digital impulse response filter measurements system uses the Median type mode filter.

The fast DIRF 'gait' object image motion measurement upon velocity by DiRM-speed of Figure 9 from Figures 1/2/3 & 7.

The use of image motion class flags [ bIlto4 ] is also extended to speed measurements by item 47 as b(11). The detection of motion class boundaries is along a frame scan line immediately previous to and then following the single interframe motion change or 'aut' condition.

The production of speed flag b(11) pulses follows from earlier b(1) pulses held in a spatial interpixel registerC17 as item block 34 to feed delayed pulses item 37 to item block 38. The concurrent 'aut' condition flag b(2) or b(3) pulses as item 18 are applied to the 2-input boolean AND gate function of item block 38. The occurrence of an output pulse as item 44 starts the measurand bitstream flow of b(11) as item 47.

Either flag b(2) or b(3) pulses as item 18 may be used to 'start' a (up-)counter using a high speed astable clock of item 39. The higher count rate is to give a magnified readout value as item 47 for ease of reading. Here the boundary edge condition between motions b(2) and b(1) has been used for 'start' b(12) signalling. The (up-)counter boolean AND gate function of item block 10 is held 'on' or 'high' by the lower (N)AND gate function of item block 34 until the boundary edge condition between motions b(2) and b(4) occurs as flag b(10).

The single interfame change state 'aut' pulses are held over the spatial interpixel period oP. in item block 45 as the register (2 ] (or register (3 ] ) delay. The boundary edge condition between flag b(2) and b(4) pulses provides the negative stop pulse dL) of b(10) as item 43. The speed measurand of b(11) of item 47 as the bitstream flow is halted to give a final digital speed readout value from item block 17 of Figure 2.

Note: - The flag b(2) pulses represent 'cessation of object motion while flag b(3) pulses would represent 'commencement' of object motion as item 18.

n selected line number N from the range or batch CNt pN2 ] and a manual disable function of item block 36 allow high frequency clock pulses through the boolean AND gate function item block 10. The higher metric resultant as item 47 is code converted for a magnified digital readout display as already detailed before and illustrated in Figures 2, 4/5/6 & 7.

Both 'DIRM - Linear' and 'DIRM - Speed' use a high frequency measurand clock to overcome the otherwise limited digital reading magnitude values, as allowed by practical digital video and television system standards, and the pixel cell scan speeds. The metric precision of the measurand bitstream 'start' and 'stop' switch timing will depend upon the method basis of the video system accuracy used.

Claims

Claims, 1. n method using a video image 4-channel motion class filter to generate flag pulses (bilto4 ] for further filtering, measurements, control or signalling from a single interframe analysis of a three frame video cascade producing intrafield temporal variables p, q for a 4-way boolean architecture pulse filter and logic. 2. n method as claimed in claim 1 to measure video flag CbIlto4 ] quantities for value display by frame scan line counting of motion class flag [ bIlto4 ] pulses or edge-transitions from one or more video frame lines. 3. n method as claimed in claim 2 to count the derived quantity of 'uncovering' motion class 2 flags from processed single intrafield image change conditions of flag b(5) to give a readout display value, signalling or control. 4. n method as claimed in claim 3 to signal the image motion class boundary condition change from 'covering' flag b(3) to a 'fully motive' flag b(4) and the state of aiming for machine-vision signalling, trigger edge firing or control at commencement of image object motion. 5. A method as claimed in claim 4 for linear dimension measurement of a scene object by counting the video image motion class flag b(1) pulses for a quantity readout display value, signalling or control. 6. A method as claimed in claim 5 for linear dimension measurement of an image object using flag (bIlto4 ] and derived pulses incorporating a high frequency clock to enhance display value magnitudes by using image boundary edge conditions b(2)/b(3) with respectively b(1) or b(4) for 'start' and 'stop' switching of the pulse counter. This by using intermediate flag b(7) 'aut' pulses to give signalling flag b(8) pulses to 'start', and flag b(9) pulses to 'stop' functions. 7. A method as claimed in claim 6 where a sampling of three binary q intrafield values, as claimed in claim 1 onwards, within an expected serial of four values are compensated for single error by producing the anticipated 'q' output binary by a three gate AND tree, across three pixel cell delay lines sP., with each gate fed into the final 3-input OR gate function to provide the output 'q' binary. 8. n method as claimed in claim 7 for enhancing image object speed measurement using flag (bIlto4 ] pulses, here incorporating a high frequency clock to enhance the display value magnitude by using image boundary edge conditions of flags b(2)/b(3) with respectively flag b(4) for the 'start' b(10) pulse, and with b(1) for the 'stop' b(12) pulse to 0laims(continued) (8) output the b(11) pulse for the quantity readout display value, signalling or control. 9. n method as claimed in claim 8 for generating the boundary condition flag b(12) for image change to or from the 'fully motive' image condition, and flag b(10) for image change to or from the 'stationary' image condition. 10. fi method as claimed in claim 9 where the boolean AND gate function output to the counter can enable and disable the count pulses for measurand control by external and/or manual means. 11. n method as claimed in claim 10 where the counter AND gate function input is enabled for a specific frame line scan period N, or line periods Ni ---# N2, to enable the measurand and control. 12. n video digital impulse response filter system of classified pulses for image object logic condition measurement, signalling and control as herewith described and illustrated by a set of drawings. 13. n method of scene object measurement, signalling and control using video image motion classification and display using motion class flag EbiItoI2 ] pulses as hereinbefore described incorporating a 4-channel input filter and logic. Amendments to the claims have been filed,

1. A method of count quantification of video image 4-channel motion class singletons to generate flag pulses CbIltol2 ] tf b(m) by neural enumeration for measurement pulses and spats~, reinforcement logic-filter operations from single At interframe analyses of a three-frame video cascade which results in temporal intrafield variables p, q as input to the 1 to 4-way boolean architecture and hamiltonian channel to feed binary transition logic.

2. 4 method as claimed in claim 1 where lateral scan sampling of three binary q intrafield values, as claimed in claim 1 onwards, within an expected serial of four values are boolean compensated for single error by producing the anticipated correct 'q' output binary by a three gate AND tree, across a triple span of three m-spatial pixel-cell delay lines ski/3 which may have differential delay, and with each neural gate pulsar fed into the final hierarchial 3-input OR gate function to reinforce the monomode output 'q' of a correction binary for a neural flag b(m) or m-descriptor pulse.

3. A method of counting as claimed in claims 1 and/or 2 to measure video flag (bilt-o4 ] quantities for value display by frame line-scan enumeration of motion class singleton tbIlto4 ] pulses or edge-transitions from one or more video frame-lines.

4. A method as claimed in claims 1 and/or 2 to count the quaternary derived quantity of 'uncovering' motion class 2 flags from processed single intrafield image change conditions of flag b(S) to give a readout display value, signalling or control.

5. A method as claimed in claim 1 and/or 2 to signal the image motion class boundary condition change from 'covering' flag b(3) to a 'fully motive' flag b(4) and the edge-state of aiming for machine-vision b(m) signalling, trigger-edge firing or digital control at the commencement boundary of image object motion.

6. A method as claimed in claims 1 andfor 2 for neural linear dimension measurement of a scene object by boolean counting the video image motion class flag b(1) pulses for enumerating a quantity readout display value, neural signalling or digital control.

7. n method of pulsing as claimed in claim 6 for linear dimension measurement of an image object using flag [ bIlto4 ] and derived pulses quantising a high frequency clock to protract display value magnitudes by using image boundary-edge conditions b(2)/b(3) with respectively b(l) or b(4) for 'start' and 'stop' switching of the pulse counter. This is performed by using intermediate flag b(7) 'aut' pulses to give signalling flag b(8) pulses to 'start', and flag b(9) pulses to 'stop' function neural executions.

claims (Continued).

8. n method of pulsing as claimed in claim 7 for enhancing image object speed measurement using neural EbIlto4 ] pulses quantising a high frequency clock to protract the display value magnitude by using image boundary-edge change conditions of flags b(2)/b(3) with neurally flag b(4) for the start b(10) pulse, and with b(1) for the 'stop b(12) pulse to output the b(11) pulse for the quantity readout display value, signalling or control from pixel changes.

9. R method as claimed in claim 8 for pixels generating the boundary conditionrflag b(12) for image change to or from the 'fully motive' image condition, and flag b(10) for image change to or from the 'stationary' image condition.

10. Q method as claimed in claim 9 where the boolean AND gate function output to the pixelisation counter can enable and disable the count pulses for measurand control by local or external logic pulse control or network and/or manual means.

11. A method as claimed in claim 10 where the counter fiND gate function input is enabled for a specific frame scan-line period N, or scan-line periods N1#N2, to enable the quantisation of neural measurands and control pulsars.

12. A device as claimed in claims 1 and 2 of a Nedian spatial video line-scan single error correction for a neural digital impulse response filter in threshold binary 'q'- > ( & .

Median transfer, Najority flag filter and/or any other monomode pulse b(m) flag location, including the m-descriptor for a DATV / DIRF system, video instrumentation, neural networking or cellular automata from pixel change conditions.

13. fi method as claimed in claim 12 of quantifying scene object measurement scanning of a digital video image motion, with (DIRF) counts or of b(m) pulses for neural count signalling, cellular automata, networking and control using video image motion classification and display (for image quaternions) to enumerate motion class flag [ bIltol2 ] pulses from a finite temporal-domain impulse response filter and a digital impulse response filter or DIRF.

14. n method as claimed in claims 1, 2, and 12 of bandwidth reduction in which for revealed occlusion error recovery a scanning suppression frame is reconstructed at source for later insertion between two wholly transmitted or recorded fields for picture display at destination. The suppression frame is transmitted along a DSTV channel along with a motion classification division pixel input to the Nedian monomode correction reinforcement of a digital impulse response filter (DIRF) in the compression tier for reconstructing the full bandwidth T.V. signal at the destination.