KR101348855B1 - Method, apparatus and image capturing system for real-time image stabilization with reduced search range using motion sensors - Google Patents

Abstract

According to the present invention, a real-time image stabilization method for imaging system including an image sensor, and a motion sensor capable of generating pitch information and yaw information comprising angle values comprises the steps of: segmenting the entire area of an image into a quadrant and previously setting four or more unit search areas including at least one of the quadrant; receiving an image from the image sensor in every frame; receiving, from the motion sensor, a motion sensor signal including at least pitch information and yaw information in every frame; selecting one of four or more unit search areas as a reduced search area, based on unit motion vector consisting of the vertical components of the pitch information and the horizontal components of the yaw information, in every frame; operating motion vector in the selected reduced search area with respect to an image of a current frame and an image of a previous frame, in every frame; and displacing the image of the current frame in a reverse direction of the motion vector operated in every frame to create an image for which the motion of a camera is compensated. [Reference numerals] (AA) Start; (BB) End; (S31) Segment the entire area of an image into a quadrant and previously set four or more unit search areas including at least one of the quadrant; (S32) Receive image signals from an image sensor; (S33) Receive, from a motion sensor, a motion sensor signal including pitch information and yaw information; (S34) Synchronize the image signals with the pitch information and the yaw information; (S35) Select one of four or more unit search areas as a reduced search area, based on unit motion vector consisting of the vertical components of the pitch information and the horizontal components of the yaw information; (S36) Operate compensated motion vector in the selected reduced search area with respect to an image of a current frame and an image of a previous frame; (S37) Displace and rotate the image according to the operated compensated motion vector

Description

Real-time image stabilization method, apparatus, and imaging device for reducing search area using motion sensor TECHNICAL FIELD AND IMAGE CAPTURING SYSTEM FOR REAL-TIME IMAGE STABILIZATION WITH REDUCED SEARCH RANGE USING MOTION SENSORS}

The present invention relates to an image stabilization technology, and more particularly, to a real-time image stabilization technology using a motion sensor.

Cameras that you hold in your hand will be affected by the slightest shaking of your hand if the shutter speed is not secured. As the resolution of the image pickup device increases, fine camera shake becomes a sensitive factor that affects the quality of the image.In this high-end camera, the image stabilization technology enables to obtain a stable image even at a relatively low shutter speed in an environment where light quantity is not secured. It was.

However, the optical image stabilization or image stabilization technology of such high-end cameras is a method of mechanically driving an imaging device or a lens to compensate for vibration, requires a special structure and high manufacturing cost, and consumes a lot of power. It is difficult to apply to surveillance cameras for cars.

Digital image stabilization technology compensates for the image itself affected by the vibration applied to the camera instead of mechanically driving the image pickup device or lens, and has no mechanical configuration added to the image pickup device or lens. Applicable to However, the real-time digital image stabilization technology processes images in real time and calculates motion vectors of pixels according to vibrations applied to the camera rather than the subject's own movement. Requires bandwidth and computing power.

An object of the present invention is to provide a real-time image stabilization method and apparatus for reducing a search area using a motion sensor.

Real time image stabilization method according to an embodiment of the present invention,

A real-time image stabilization method for an imaging system including an image sensor and a motion sensor capable of generating pitch information and yaw information having an angle value,

Dividing the entire area of the image into quadrants and presetting at least four unit search regions including at least one of the quadrants;

Receiving an image from the image sensor every frame;

Receiving a motion sensor signal including at least pitch information and yaw information from the motion sensor every frame;

Selecting one unit search area from among the at least four unit search areas as a reduced search area based on a unit motion vector including a vertical component of the pitch information and a horizontal component of the yaw information every frame;

For each frame, calculating a correction motion vector comprising displacement components and rotational components in the selected reduction search region, with respect to the image of the current frame and the image of the previous frame; And

The method may include generating an image stabilized correction image by displacing and rotating the image of the current frame according to the calculated displacement component and rotation component of the corrected motion vector every frame.

In at least one example embodiment, the at least four unit search areas may be set equal to each quadrant of the entire area of the image.

In at least one example embodiment, the at least four unit search areas may be set to be equal to each of the areas further included in each quadrant of the entire area of the image by a predetermined margin width at the boundary of the quadrant.

In one embodiment, the predetermined margin width may be determined from a value greater than 0% and less than 50% of the resolution of the entire area of the image.

In one embodiment, the real-time image stabilization method

The method may further include synchronizing the image signals and the motion sensor signal every frame.

In one embodiment, the real-time image stabilization method

Comparing the magnitudes of the calculated displacement components or rotational components of the corrected motion vector to predetermined thresholds,

If the size of one of the components of the calculated corrected motion vector exceeds predetermined thresholds, the calculated corrected motion vector may be discarded, otherwise the process may proceed to generating a corrected image.

According to another aspect of the present invention, a real-time image stabilization apparatus for an imaging system including an image sensor and a motion sensor capable of generating pitch information and yaw information having angle values,

A unit consisting of a vertical component of pitch information and a horizontal component of yaw information that are divided into quadrants and have at least four unit search regions including at least one of the quadrants set in advance, and input every frame A search region reduction unit selecting one unit search region from among the at least four unit search regions as a reduction search region based on a motion vector;

A motion vector calculator for calculating a corrected motion vector including a displacement component and a rotation component in the selected reduction search region for each image, with respect to an image of a current frame and an image of a previous frame; And

Each frame may include a motion compensator for generating an image stabilized correction image by displacing and rotating the image of the current frame according to the calculated displacement component and rotation component of the corrected motion vector.

In at least one example embodiment, the at least four unit search areas may be set equal to each quadrant of the entire area of the image.

In at least one example embodiment, the at least four unit search areas may be set to be equal to each of the areas further included in each quadrant of the entire area of the image by a predetermined margin width at the boundary of the quadrant.

In one embodiment, the predetermined margin width may be determined from a value greater than 0% and less than 50% of the resolution of the entire area of the image.

In one embodiment, the real-time image stabilization device

The apparatus may further include a synchronizer configured to synchronize an image input from the image sensor every frame and a motion sensor signal including at least pitch information and yaw information input from the motion sensor every frame.

In one embodiment, the motion compensation unit,

The magnitude of the displacement component or rotational component of the calculated corrected motion vector is compared to predetermined thresholds, and if the magnitude of one of the components of the calculated corrected motion vector exceeds predetermined thresholds, Discard the corrected motion vector, otherwise generate a corrected image.

An imaging system according to another aspect of the present invention,

Image sensor;

A motion sensor capable of generating pitch information and yaw information in angular values; And

Receives image signals from the image sensor every frame and pitch information and yaw information from the motion sensor, divides the entire area of the image into quadrants, and presets at least four unit search regions including at least one of the quadrants. In each frame, one unit search area is selected as the reduced search area based on a unit motion vector including a vertical component of the pitch information and a horizontal component of the yaw information. And a correction motion vector including a displacement component and a rotation component in the selected reduction search region with respect to an image of a current frame and an image of a previous frame in the selected reduction search region every frame, and then calculate the calculated motion frame every frame. Displacement component of the compensation motion vector, and Displacement and rotating the image of the current frame in accordance with the Ingredients to may include a real-time image stabilization apparatus for generating a corrected video image stabilization.

In at least one example embodiment, the at least four unit search areas may be set equal to each quadrant of the entire area of the image.

In at least one example embodiment, the at least four unit search areas may be set to be equal to each of the areas further included in each quadrant of the entire area of the image by a predetermined margin width at the boundary of the quadrant.

In one embodiment, the predetermined margin width may be determined from a value greater than 0% and less than 50% of the resolution of the entire area of the image.

In one embodiment, the real-time image stabilization device,

It is operable to synchronize an image input from the image sensor every frame and a motion sensor signal including at least pitch information and yaw information input from the motion sensor every frame.

In one embodiment, the real-time image stabilization device,

The magnitude of the displacement component or rotational component of the calculated corrected motion vector is compared to predetermined thresholds, and if the magnitude of one of the components of the calculated corrected motion vector exceeds predetermined thresholds, Discard the corrected motion vector, otherwise generate a corrected image.

According to the real-time image stabilization method and apparatus of the present invention, image stabilization can be performed in real time while reducing computational power and power consumption by reducing the search area using a motion sensor.

1 is a conceptual diagram illustrating a relationship between a camera movement and reduction of a search area in a real time image stabilization method according to an embodiment of the present invention.
2 is a conceptual diagram illustrating a relationship between up, down, left, and right motion components and a reduction search region detected by a motion sensor of a camera in a real-time image stabilization method according to an embodiment of the present invention.
3 is a flowchart illustrating a real-time image stabilization method according to an embodiment of the present invention.
4 is a block diagram illustrating a real-time image stabilization apparatus and an imaging system according to an embodiment of the present invention.

For the embodiments of the invention disclosed herein, specific structural and functional descriptions are set forth for the purpose of describing an embodiment of the invention only, and it is to be understood that the embodiments of the invention may be practiced in various forms, The present invention should not be construed as limited to the embodiments described in Figs.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. The same reference numerals are used for the same constituent elements in the drawings and redundant explanations for the same constituent elements are omitted.

1 is a conceptual diagram illustrating a relationship between a camera movement and reduction of a search area in a real time image stabilization method according to an embodiment of the present invention.

Referring to FIG. 1, the movement of the camera due to an unwanted external force applied from the outside, such as the shaking of a human hand or the shaking of a mount, may include pitch as a vertical rotation component, yaw as a horizontal rotation component, and in-situ rotation. It can be expressed as a roll, which is a component.

As the camera moves, the field of view of the camera changes, and the subject moves opposite to the camera movement in the captured image. When comparing the image taken before the camera moves with the image taken after the camera moves, the common part of the two images is reduced compared to the image before the camera moves. In order to accurately calculate the movement of the camera, a motion vector must be computed between the images before and after the camera moves. In order to do this, a motion vector calculation for an object determined to be the same in both images is required.

Since such objects will be photographed within a common viewing angle both before and after the camera moves, the computation of the motion vector is substantially the area corresponding to the common viewing angle before and after the camera moves. It can be concluded that it can be done only in.

However, since the movement of the camera related to image stabilization is the level of unconscious hand shake or vibration of the mount, even if the motion vector search region is limited to a region corresponding to a common viewing angle before and after the movement of the camera, it is difficult to obtain a great advantage. On the contrary, if the degree of vibration differs from frame to frame, the viewing angle common to before and after the movement of the camera changes, and the size of the motion vector search region also changes.

Therefore, there is a need for a more aggressive and fixed search area reduction method.

2 is a conceptual diagram illustrating a relationship between up, down, left, and right motion components and a reduction search region detected by a motion sensor of a camera in a real-time image stabilization method according to an embodiment of the present invention.

The motion sensor mountable to the camera may provide pitch, yaw and roll information. The pitch, yaw and roll information provided as an angular value may not be utilized as it is in the digital image stabilization method. The digital image stabilization method is a method of obtaining a two-dimensional translation of an image due to camera movement and correcting the influence of the camera movement based on this. Since the pitch, yaw and roll information are all rotation angles, pitch, In order to obtain the vertical or horizontal displacement of the subject in the imaging surface from the yaw and roll information, a separate complicated calculation is required.

In the present invention, instead of performing a complicated operation with the angle values of the pitch, yaw and roll information in this way, in advance using the vertical and horizontal components of the object motion as indicated by the sign of the angle values of the pitch and yaw information, One of the defined small search areas can be selected.

In the present invention, the camera is equipped with a three-dimensional motion sensor, the motion sensor of the camera may output the pitch, yaw and roll information in at least one form of angle, angular velocity or angular acceleration. At this time, although the motion sensing principle itself is performed based on the angular acceleration, it must be converted into an angular value in order to be finally used as pitch, yaw and roll information.

Specifically, referring to FIG. 2, assuming that the camera moves rapidly downward to the left due to disturbance, the subject in a stationary state is based on the subject photographing position before the camera moves, as opposed to the movement of the camera. When done, it will be captured as if it has moved to the upper right corner.

In this case, instead of obtaining a motion vector directly from the angle, angular velocity, or angular acceleration values generated by the motion sensor, the present invention uses only up, down, left, and right direction components according to the angle value output as pitch information and yaw information. However, among the pitch, yaw, and roll information output from the motion sensor of the camera, the roll information may be ignored because it is almost independent of the two-dimensional displacement of the image due to the movement of the camera.

In other words, in the present invention, the pitch information identifies the vertical direction component of the subject movement, that is, whether it is upward or downward, and the yaw information identifies the horizontal direction component of the subject movement, that is, whether it is left or right, and each determined vertical And a horizontal motion component to determine the unit motion vector u.

The degree of disturbance of the subject varies depending on the degree of disturbance applied to the camera, but the movement of the objects A and A 'of the subject between the image 21 before the camera shakes and the image 22 after the camera shakes. It has the same direction as the unit motion vector u.

The unit motion vector u may be four vectors: right upward, right downward, left upward, or left downward. In this case, when the angle value is extremely 0, the default classification may be upward or leftward.

Therefore, since the directionality of the motion vector is reliably determined before calculating the magnitude of the motion vector of the subject due to the movement of the camera, the reduction of the calculation region itself for calculating the motion vector does not affect the accuracy.

In Fig. 2, when calculating the motion vector between the objects A and A 'of the main subject from the images 21 and 22 before and after the camera shake, the inventor shakes the subject object A' after shaking. It is expected that the result calculated on the basis of only the upper right quadrant Q1, in which the difference of the previous object object A is most apparent, will not be significantly different from the result calculated on the basis of the entire image.

However, even when the degree of movement is extremely small, the accuracy may be lowered when the calculation is performed in one of the quadrants Q1 to Q4, so that a predetermined margin width, for example, the full resolution, is defined at the boundary of each quadrant Q1 to Q4. An area containing 10% or more may be set as a reduced search area.

Accordingly, the present invention presets four unit search areas in the entire area of the image, and every frame, among the four unit search areas, the main subject objects of the images 21 and 22 before and after the shaking of the camera ( One unit search region 23 suitable for calculating a motion vector between A and A 'may be selected as the reduced search region.

In this case, the reduced search region may be selected as the unit search region 23 that matches the direction of the unit motion vector u.

If the unit search area 23 is set to one of the quadrants, the calculation amount for calculating the motion vector is reduced to 0.5 * 0.5 = 0.25 compared with the case of obtaining the motion vector based on the entire area of the image. Reduced by% If the unit search area 23 is set to one of the quadrants and an area further including a margin of 10% of the full resolution in the quadrant, the calculation amount for calculating the motion vector is reduced to 0.6 * 0.6 = 0.36. This reduces by 64%.

At this time, the width of the margin may be appropriately selected according to the trade-off relationship between accuracy and speed between values greater than 0% and less than 50% of the resolution.

3 is a flowchart illustrating a real-time image stabilization method according to an embodiment of the present invention.

Referring to FIG. 3, in step S31, the real-time image stabilization method divides the entire region of the image into quadrants and presets at least four unit search regions including at least one of the quadrants.

According to an embodiment, at least four unit search areas may be set to be equal to each quadrant of the entire area of the image.

In another embodiment, the at least four unit search regions may be set to be equal to each of the regions further included in each quadrant of the entire region of the image by a predetermined margin width at the boundary of the quadrant.

In this case, the predetermined margin width may be selected according to the design from a value greater than 0% and less than 50% of the full area resolution of the image, and preferably 10% of the full area resolution of the image.

In step S32, image signals are input from the image sensor every frame, and in step S33, a motion sensor signal including at least pitch information and yaw information is received from the motion sensor every frame.

In step S34, the image signals and the pitch information and the yaw information are synchronized every frame.

In step S35, one unit search area is selected from among at least four unit search areas as a reduced search area based on a unit motion vector including a vertical component of pitch information and a horizontal component of yaw information every frame. do.

In step S36, a correction motion vector including a displacement component and a rotation component in the selected reduction search region is calculated for each image of the current frame and the previous frame.

At this time, according to the embodiment, each component value of the corrected motion vector calculated in step S36 is compared to a predetermined threshold, and if the magnitude of the calculated corrected motion vector component exceeds the predetermined threshold, Since the motion vector may be due to the camera's intentional change of view angle or large movement of the subject itself, not due to the accidental vibration of the camera, the result of calculating the motion vector calculation may be discarded without proceeding to the correction procedure of the next step (S37). It may be.

In step S37, the image is stabilized by displacing and rotating the image according to each component of the corrected motion vector calculated every frame.

4 is a block diagram illustrating a real-time image stabilization apparatus and an imaging system according to an embodiment of the present invention.

Referring to FIG. 4, the imaging system 40 may include a real time image stabilization device 41, an image sensor 42, a motion sensor 43, and a post processor 44.

The real-time image stabilization apparatus 41 divides the entire area of the image into quadrants and presets at least four unit search regions including at least one of the quadrants.

In addition, the real-time image stabilization device 41 receives an image signal from the image sensor 42 every frame, and receives a motion sensor signal including at least pitch information and yaw information from the motion sensor 43.

Subsequently, the real-time image stabilization device 41 synchronizes the image signals with the pitch information and the yaw information every frame, and based on the unit motion vector composed of the vertical component of the pitch information and the horizontal component of the yaw information every frame. To select one unit search area from among at least four unit search areas as a reduction search area, and include a displacement component and a rotational component with respect to the image of the current frame and the image of the previous frame in the selected reduction search area every frame. After calculating the motion vector, a corrected image is generated by displacing and rotating the image of the current frame according to the displacement component and the rotation component of the corrected motion vector calculated every frame.

To this end, the real-time image stabilization device 41 may include a synchronizer 411, a buffer 412, a search region reducer 413, a motion vector calculator 414, and a motion compensator 415.

The synchronizer 411 receives image signals from an image sensor every frame and a motion sensor signal including at least pitch information and yaw information from a motion sensor, and receives image signals, pitch information, and yaw information every frame. Synchronize.

The buffer 412 may buffer the synchronized image signals and the motion sensor signals prior to the operation for motion compensation.

The search area reducer 413 divides the entire area of the image into quadrants and sets at least four unit search areas including at least one of the quadrants in advance, and then, after each frame, the search area reducer 413 Based on the unit motion vector including the vertical component and the horizontal component of the yaw information, one unit search region among at least four unit search regions may be selected as the reduced search region.

In this case, according to an embodiment, at least four unit search regions may be set to each of four quadrants.

In another embodiment, the at least four unit search regions may be set such that each of the four quadrants further includes a predetermined margin width at the boundary of the quadrant.

In this case, the predetermined margin width may be selected according to the design from a value greater than 0% and less than 50% of the image resolution.

The motion vector calculator 414 calculates a correction motion vector including displacement components and rotation components in the reduced search region selected by the search region reducer 413 instead of the entire region of the image with respect to the image of the current frame and the previous frame. do.

The motion compensator 415 generates and outputs a stabilized corrected image by displacing and rotating the image of the current frame according to the displacement component and the rotation component of the corrected motion vector calculated by the motion vector calculator 414.

The post processor 44 may store, compress, convert, or display the stabilized image output from the real time image stabilization device 41.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, Modification is possible. Accordingly, the spirit of the present invention should be understood only in accordance with the following claims, and all of the equivalent or equivalent variations will fall within the scope of the present invention.

Further, the apparatus according to the present invention can be embodied as a code that can be read by the information processing apparatus on a recording medium that can be read by the information processing apparatus. The recording medium readable by the information processing apparatus includes all kinds of recording apparatuses in which data that can be read by the information processing system is stored. Examples of the recording medium include a ROM, a RAM, an optical disk, a magnetic tape, a floppy disk, a hard disk, a nonvolatile memory, and the like, and a carrier wave (for example, transmission via the Internet).

40 Imaging System 41 Real-time Image Stabilizer
411 Synchronizer 412 Buffer
413 Navigation Area Reduction Unit 414 Motion Vector Computation Unit
415 Motion Compensator 42 Image Sensor
43 Motion Sensor 44 Post Processing Unit

Claims (19)

A real-time image stabilization method for an imaging system including an image sensor and a motion sensor capable of generating pitch information and yaw information having an angle value,
Dividing the entire area of the image into quadrants and presetting at least four unit search regions including at least one of the quadrants;
Receiving an image from the image sensor every frame;
Receiving a motion sensor signal including at least pitch information and yaw information from the motion sensor every frame;
Selecting one unit search area from among the at least four unit search areas as a reduced search area based on a unit motion vector including a vertical component of the pitch information and a horizontal component of the yaw information every frame;
For each frame, calculating a correction motion vector comprising displacement components and rotational components in the selected reduction search region, with respect to the image of the current frame and the image of the previous frame; And
And displacing and rotating the image of the current frame according to the calculated displacement component and rotation component of the corrected motion vector every frame to generate an image stabilized corrected image. The method according to claim 1, wherein the at least four unit search areas are set equal to each quadrant of the entire area of the image. The real-time image of claim 1, wherein the at least four unit search regions are set to be equal to each of the regions further included in each quadrant of the entire region of the image by a predetermined margin width at the boundary of the quadrant. Stabilization method. 4. The method of claim 3, wherein the predetermined margin width is determined from a value greater than 0% and less than 50% of the resolution of the entire area of the image. The method according to claim 1,
And synchronizing the video signals and the motion sensor signal every frame. The method of claim 1, further comprising comparing magnitudes of the calculated displacement component or rotational component of the corrected motion vector to predetermined thresholds,
If the magnitude of one of the components of the calculated corrected motion vector exceeds predetermined thresholds, the calculated corrected motion vector is discarded; otherwise, the process proceeds to generating a corrected image. Image stabilization method. Information for implementing a real-time image stabilization method for an imaging system including an image sensor according to any one of claims 1 to 6 in the information processing apparatus and a motion sensor capable of generating pitch information and yaw information having angle values. A recording medium containing a program that can be read by a processing device. A real-time image stabilization apparatus for an imaging system including an image sensor and a motion sensor capable of generating pitch information and yaw information having an angle value,
A unit consisting of a vertical component of pitch information and a horizontal component of yaw information that are divided into quadrants and have at least four unit search regions including at least one of the quadrants set in advance, and input every frame A search region reduction unit selecting one unit search region from among the at least four unit search regions as a reduction search region based on a motion vector;
A motion vector calculator for calculating a corrected motion vector including a displacement component and a rotation component in the selected reduction search region for each image, with respect to an image of a current frame and an image of a previous frame; And
And a motion compensator for generating an image stabilized corrected image by displacing and rotating the image of the current frame according to the calculated displacement component and rotation component of the corrected motion vector every frame. The apparatus of claim 8, wherein the at least four unit search regions are set equal to each quadrant of the entire region of the image. The real-time image of claim 8, wherein the at least four unit search regions are set to be equal to each of the regions further included in each quadrant of the entire region of the image by a predetermined margin width at the boundary of the quadrant. Stabilization device. The apparatus of claim 10, wherein the predetermined margin width is determined from a value greater than 0% and less than 50% of the resolution of the entire area of the image. The method according to claim 8,
And a synchronization unit for synchronizing an image input from the image sensor every frame and a motion sensor signal including at least pitch information and yaw information input from the motion sensor every frame. The method of claim 8, wherein the motion compensation unit,
The magnitude of the displacement component or rotational component of the calculated corrected motion vector is compared to predetermined thresholds, and if the magnitude of one of the components of the calculated corrected motion vector exceeds predetermined thresholds, Discarding the corrected motion vector, otherwise generating a corrected image. Image sensor;
A motion sensor capable of generating pitch information and yaw information in angular values; And
Receives image signals from the image sensor every frame and pitch information and yaw information from the motion sensor, divides the entire area of the image into quadrants, and presets at least four unit search regions including at least one of the quadrants. In each frame, one unit search area is selected as the reduced search area based on a unit motion vector including a vertical component of the pitch information and a horizontal component of the yaw information. And a correction motion vector including a displacement component and a rotation component in the selected reduction search region with respect to an image of a current frame and an image of a previous frame in the selected reduction search region every frame, and then calculate the calculated motion frame every frame. Displacement component of the compensation motion vector, and By the displacement and rotating the image of the current frame according to the former components including real-time image stabilization apparatus for generating a corrected image stabilization video imaging system according to claim. The imaging system of claim 14, wherein the at least four unit search areas are set equal to each quadrant of the entire area of the image. 15. The imaging system of claim 14, wherein the at least four unit search areas are set equal to each of the areas further included in each quadrant of the entire area of the image by a predetermined margin width at the boundary of the quadrant. . 17. The imaging system of claim 16, wherein the predetermined margin width is determined from a value greater than 0% and less than 50% of the full area resolution of the image. The apparatus of claim 14, wherein the real-time image stabilization device is
And a motion sensor signal including an image input from the image sensor every frame and at least pitch information and yaw information input from the motion sensor every frame. The apparatus of claim 14, wherein the real-time image stabilization device is
The magnitude of the displacement component or rotational component of the calculated corrected motion vector is compared to predetermined thresholds, and if the magnitude of one of the components of the calculated corrected motion vector exceeds predetermined thresholds, Discarding the corrected motion vector, otherwise generating a corrected image.

Images