Image analysis – Applications – Motion or velocity measuring
Reexamination Certificate
2001-01-25
2004-02-03
Johnson, Timothy M. (Department: 2625)
Image analysis
Applications
Motion or velocity measuring
C382S219000, C382S224000, C382S236000, C375S240160, C348S699000
Reexamination Certificate
active
06687388
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a picture processing apparatus for detecting a moving vector of a moving picture.
2. Description of the Related Art
To eliminate noise from a picture, a motion adaptive type recursive filter has been used. An example of the structure of the motion adaptor recursive filter is shown in FIG.
1
. Picture data of a considered frame (namely, an input frame at each time point) is supplied to a subtracting circuit
1100
. Picture data of the preceding frame (one frame earlier than the considered frame) is supplied from a frame memory
1099
to the subtracting circuit
1100
. An output of the subtracting circuit
1100
is supplied to an absolute value calculating circuit
1101
. The absolute value calculating circuit
1101
converts the output data of the subtracting circuit
1100
into an absolute value. The absolute value is supplied to a threshold value processing circuit
1102
. The threshold value processing circuit
1102
compares the supplied absolute value with a predetermined threshold value so as to determine whether the picture moves or stops for each pixel.
The determined result of the threshold value processing circuit
1102
is supplied to a weight designated value generating circuit
1103
. The weight designated value generating circuit
1103
designates a weight designated value k corresponding to the supplied determined result. The weight designated value k is supplied to an amplifier
1105
. The amplifier
1105
adjusts the amplitude of the input frame. An amplifier
1104
amplifies the input signal by k times. On the other hand, the amplifier
1105
amplifies pixel date stored in the frame memory by (1−k) times.
When the threshold value processing circuit
1102
has determined that the picture of the current frame stops, a fixed value ranging from 0 to 0.5 as the value k is designated. Thus, an output of an adder
1106
disposed downstream of the amplifiers
1104
and
1105
becomes a weighted value of a pixel value of the considered frame and a pixel value of the preceding frame. On the other hand, when the threshold value processing circuit
1102
has determined that the picture of the current frame moves, “1” is designated to the value k. As a result, the adder
1106
directly outputs a pixel value of the considered frame.
The above-described motion adaptive type recursive filter has problems (a) to (d) that follow. (a) Since a still portion is weighted with the fixed value k, if the intensity and/or direction of noise varies, the noise cannot be adequately eliminated. (b) When the noise level is large, a still portion is mistakenly determined as a moving picture. Thus, the noise eliminating effect deteriorates. (c) A moving portion may be mistakenly determined as a still picture. As a result, deterioration such as an occurrence of a trailing shape portion may take place. (d) Noise cannot be eliminated from a moving portion.
In addition, a noise eliminating method using a class category adaptive process has been proposed as Japanese Patent Application No. 10-304058. In the method, a block matching operation is performed so as to detect a moving vector. Corresponding to the detected moving vector, a class category adaptive process using pixels at the same position on frames is performed. Although the problems of the motion adaptive type recursive filter can be solved or alleviated using the method, it has the following problems.
(e) Since the block matching operation is performed, a heavy load is imposed to the apparatus. (f) The block matching operation should be performed between two adjacent frames (for example, between the frames t and (t+1), between the frames (t+1) and (t+2), and so forth). If the block matching operation fails for at least two adjacent frames, the noise elimination effect deteriorates. (g) In the block matching operation, a motion of a small object in a block cannot be detected. Thus, such a motion may result in an inadequate process.
In addition, as other related art references rather than the motion adaptive type recursive filter for detecting a moving vector, block matching method, slope method, and so forth are known. In the block matching method, the sum of differences of pixels of each block in each searching range should be calculated and compared. Thus, the amount of calculation becomes large. Consequently, there are problems from view points of the circuit scale, the process time, and so forth. In addition, a small motion in a block cannot be accurately detected. In the slope mode, if a picture contains noise, the detection accuracy of the moving amount remarkably deteriorates.
Next, the above-described related art reference (Japanese Patent Application No. 10-304058 and corresponding U.S. patent application Ser. No. 09/602,934 filed on Oct. 25, 1999) that had not been disclosed on the filing date of the prior Japanese application of the present patent application will be described in detail.
FIG. 2
shows an example of the structure of the related art reference. As was described above, in the related art reference, in a combination of the block matching operation and the class category adaptive process, noise is eliminated from a moving picture. An input picture signal from which noise should be removed is supplied to a frame memory
200
. Frame memory
201
and
202
are connected in a shift register shape to the frame memory
200
. Thus, picture data of three chronologically successive frames are stored to the frame memories
200
,
201
, and
202
.
The frame memories
200
and
201
supply stored frames to a moving vector detecting portion
203
. The moving vector detecting portion
203
performs a block matching operation for the two supplied frames and detects a moving vector. The moving vector detecting portion
203
supplies the detected moving vector to area extracting portions
205
and
206
. In addition, the frame memories
200
,
201
, and
202
supply their stored frames to the area extracting portion
205
and
206
. A moving vector detecting portion
204
performs the block matching operation corresponding to the supplied two frames and detects a moving vector. The moving vector detecting portion
204
supplies the detected moving vector to the area extracting portions
205
and
206
.
The frames stored in the frame memories
200
,
201
,
202
are supplied to the area extracting portion
206
. The area extracting portion
206
references the moving vector supplied from the moving vector detecting portion
203
and the moving vector supplied from the moving vector detecting portion
204
and extracts a picture area at a predetermined position from the supplied frame and supplies data of the extracted picture area to a class code generating portion
207
. The class code generating portion
207
extracts a space-time pattern of pixel values corresponding to for example ADRC (Adaptive Dynamic Range Coding) method and generates a class code that represents a class categorized corresponding to the extracted pattern.
Since the picture area that is extracted by the area extracting portion
206
is used for a class category process, such a picture area is referred to as class tap. The class code is supplied to a coefficient memory
208
. The coefficient memory
208
stores a predictive coefficient set for individual classes. The coefficient memory
208
outputs a predictive coefficient corresponding to a class code. An output of the coefficient memory
208
is supplied to an estimation calculating portion
209
.
On the other hand, picture data of frames stored in the frame memories
200
,
201
, and
202
are supplied to the area extracting portion
205
. The area extracting portion
205
references moving vectors supplied from the moving vector detecting portions
204
and
203
, extracts picture areas from the picture data of the supplied frames, and supplies data of the extracted picture areas to the estimation calculating portion
209
. The estimation calculating portion
209
performs a predetermined calculatio
Kondo Tetsujiro
Nakaya Hideo
Shiraki Hisakazu
Tange Akira
Choobin Barry
Frommer William S.
Frommer & Lawrence & Haug LLP
Johnson Timothy M.
Sony Corporation
LandOfFree
Picture processing apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Picture processing apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Picture processing apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3330427