JP6361166B2 - Projector and image processing method in projector - Google Patents

Description

  The present invention relates to a projector, and more particularly to image processing including keystone correction.

  When projecting onto a screen from a projection device such as a projector, trapezoidal distortion may occur in an image projected on the screen due to the relative positional relationship between the projector and the screen. In such a case, a trapezoid correction technique that corrects trapezoidal distortion by interpolating image data so that the image projected on the screen is projected in a state close to the original image is used (for example, Patent Document 1).

  Since trapezoidal correction involves reduction and deformation, the resolution is lowered, and there are cases where the frequency band inherent to the image signal cannot be expressed. In this case, if the resolution difference between the input image and the output image is taken into consideration and the frequency band of the interpolation filter is not limited, noise such as moire may occur. On the other hand, it has been proposed to apply a sufficiently strong low-pass filter (frequency band restriction) so that moire does not occur (Patent Document 2).

JP 2002-247614 A JP 2001-78039 A

  However, if the low-pass filter is applied strongly, the image blur becomes large. In particular, when a strong low-pass filter is applied when the keystone distortion is small and the keystone correction is applied only slightly from the uncorrected state, there is a problem that the blur becomes excessively large and the image becomes difficult to see for the user.

  SUMMARY An advantage of some aspects of the invention is to solve at least a part of the problems described above, and the invention can be implemented as the following forms or application examples.

  (1) According to one aspect of the invention, a projector is provided. The projector generates a second image obtained by deforming the first image by interpolation processing using an interpolation filter in order to perform trapezoidal correction of the image projected on the projection surface according to the first image. A trapezoidal correction unit; a light modulation device that emits light representing the second image; and a projection optical system that projects light from the light modulation device onto the projection surface. A reduction rate when transforming from the first image to the second image at a plurality of predetermined points in the first image is calculated, and the band limitation of the interpolation filter is increased as the minimum value of the reduction rate is larger. The interpolation filter is adjusted so as to weaken. According to this type of projector, it is possible to make blurring inconspicuous when applying keystone correction and to suppress moire.

  (2) In the projector of the above aspect, the predetermined plurality of points may be four vertices of the first image. In the trapezoidal correction, the reduction ratio is minimized at any of the four vertices of the first image. Therefore, the reduction ratio can be easily set by setting a predetermined plurality of points as the four vertices of the first image. The minimum value can be obtained.

  (3) In the projector of the above aspect, when the minimum value of the reduction ratio is equal to or larger than a predetermined threshold, the ratio of the cutoff frequency of the interpolation filter to the reduction ratio increases as the minimum value of the reduction ratio increases. When the band limitation is weakened by increasing the adjustment coefficient represented by the following formula, and the minimum value of the reduction ratio is less than the threshold, the adjustment coefficient may be a predetermined value. According to this type of projector, when the minimum value of the reduction ratio is large and the deformation in the trapezoidal correction is small, moire is unlikely to occur, so the blurring when applying the keystone correction becomes inconspicuous by reducing the band limitation. Priority can be given to doing.

  (4) In the projector of the above aspect, the threshold value may be set to a different value for a plurality of projection modes of the projector. According to the projector of this aspect, it is possible to optimally adjust blur and moire for each projection mode.

  (5) According to one aspect of the invention, an image processing method in a projector is provided. In this image processing method, (a) a second image obtained by deforming the first image is subjected to an interpolation filter in order to perform trapezoidal correction of the image projected on the projection surface according to the first image. A step of generating by interpolation processing, (b) a step of emitting light representing the second image from the light modulation device, and (c) a step of projecting light from the light modulation device onto the projection plane. The step (a) includes: (a-1) calculating a reduction ratio when the first image is transformed into the second image at a plurality of predetermined points in the first image; (A-2) adjusting the interpolation filter so that the band limitation of the interpolation filter is weakened as the minimum value of the reduction ratio increases. According to this form of the image processing method, when the minimum value of the reduction ratio is large and the deformation in the trapezoid correction is small, moire is unlikely to occur, so blurring at the time of applying the keystone correction is reduced by weakening the band limitation. Priority can be given to making it inconspicuous.

  (6) In the image processing method of the above aspect, the predetermined plurality of points may be four vertices of the first image. In the trapezoidal correction, the reduction ratio is minimized at any of the four vertices of the first image. Therefore, the reduction ratio can be easily set by setting a predetermined plurality of points as the four vertices of the first image. The minimum value can be obtained.

  (7) In the step (a-2) of the image processing method of the above aspect, when the minimum value of the reduction ratio is equal to or greater than a predetermined threshold, the larger the minimum value of the reduction ratio, the larger the reduction ratio. When the adjustment factor represented by the ratio of the cutoff frequency of the interpolation filter is increased to weaken the band limitation, and the minimum value of the reduction ratio is less than the threshold value, the adjustment factor is a predetermined value. It may be said. According to this form of the image processing method, when the minimum value of the reduction ratio is large and the deformation in the trapezoid correction is small, moire is unlikely to occur, so blurring at the time of applying the keystone correction is reduced by weakening the band limitation. Priority can be given to making it inconspicuous.

  (8) In the image processing method of the above aspect, the threshold value may be set to a different value for a plurality of projection modes of the projector. According to this form of the image processing method, it is possible to optimally adjust blur and moire for each projection mode.

  The present invention can be realized in various forms, for example, in various forms such as an image processing method in a projector in addition to a projector.

It is explanatory drawing which shows the projector and screen of this embodiment. It is explanatory drawing which shows the relationship between a pixel and a normalization frequency. It is a graph which shows the filter characteristic of an ideal interpolation filter. It is explanatory drawing which shows an example of a specific filter coefficient. It is explanatory drawing which shows the filter characteristic of the interpolation filter which has a filter coefficient of FIG. FIG. 5 is an explanatory diagram illustrating an example of a filter coefficient used when band limitation is applied more strongly than that illustrated in FIG. It is explanatory drawing which shows the filter characteristic of the interpolation filter which has a filter coefficient of FIG. It is explanatory drawing which shows an example of the frequency characteristic of the image input into a projector. It is a block diagram of the projector in this embodiment. It is explanatory drawing which shows how to obtain | require the reduction rate (epsilon). It is a flowchart of filter coefficient calculation. It is explanatory drawing which shows the relationship between minimum reduction rate (epsilon) min and cutoff frequency adjustment coefficient (alpha).

  FIG. 1 is an explanatory diagram showing a projector and a screen (projection surface) according to this embodiment. In the present embodiment, an image (or video) is projected (projected) from the projector 10 onto the screen 20. At this time, depending on the relative positional relationship between the projector 10 and the screen 20, the image Pic1 (projected image) projected on the screen 20 may be distorted into a trapezoid, a parallelogram, or other quadrangle. Such distortion is called “trapezoidal distortion”. In such a case, an image distorted in the opposite direction with respect to the projected image on the projection surface (hereinafter also referred to as a corrected image) is generated on a light modulation device such as a liquid crystal panel in the projector 10 and the image is generated. The trapezoidal distortion is made inconspicuous by projecting on the screen 20 as Pic2. This deformation process is called trapezoidal correction and is performed using perspective transformation (projective transformation).

  When the corrected image is created in the light modulation device, for example, an interpolation process is performed. When performing trapezoidal correction by interpolation processing, it is preferable to consider the frequency band that the pre-correction image originally has. In particular, when the resolution decreases, the corrected image after trapezoidal correction may not be able to express the frequency band of the image before correction. In this case, if the frequency band of the interpolation filter is not limited, noise such as moire may occur. Moire means a striped pattern that visually occurs due to a shift in the period when a plurality of regularly repeated patterns are superimposed. In the trapezoidal correction, as shown in FIG. 1, since the reduction rate (resolution) varies depending on the position on the image, it is preferable to change the band limitation (cutoff frequency) of the interpolation filter according to the position on the image.

  FIG. 2 is an explanatory diagram showing the relationship between pixels and normalized frequencies. In a display device such as the projector 10, it is preferable to use a normalized frequency because the frequency becomes a problem in relation to the pixel. In FIG. 2, the pixel value of the hatched pixel is [1], and the pixel value of the non-hatched pixel is [0]. In general, the normalized frequency is defined by 1 / (number of pixels in one cycle). For example, when [1] and [0] are alternately repeated for each pixel, one cycle is 2 pixels and the normalized frequency is 0.5. When [1] and [0] are alternately repeated every 5 pixels, one cycle is 10 pixels and the normalized frequency is 0.1.

  When the normalized frequency is used, the frequency band that can represent the input image is 0.5 or less. Since the resolution of each image portion of the image after the keystone correction is lower than that of the original image, it is preferable to limit the frequency band according to the reduction rate (reduction rate). For example, it is preferable to apply an interpolation filter that limits the bandwidth to ½ of the frequency band of the input image in a portion where the resolution becomes ½ (reduction ratio 0.5) by the keystone correction.

  FIG. 3 is a graph showing filter characteristics of an ideal interpolation filter. In an ideal interpolation filter, when the reduction ratio ε is 0.9, the gain is 0 when the normalized frequency is 0.45 or more, and the gain is 1.0 when the normalized frequency is less than 0.45. When the reduction ratio ε is 0.8, the normalization frequency is 0.4 or more and the gain is 0, and the normalization frequency is less than 0.4 and the gain is 1.0. That is, the filter characteristic of an ideal interpolation filter is 0 when the normalized frequency is 1/2 or more of the reduction rate ε, and 1 when the normalized frequency is less than 1/2 of the reduction rate ε. .0. Thus, the ideal interpolation filter cut-off frequency Fc is equal to ½ of the reduction ratio ε.

  FIG. 4 is an explanatory diagram showing an example of specific filter coefficients. The filter coefficient is given for each reduction rate ε, and is selected according to the reduction rate ε. FIG. 4 shows filter coefficients for 10 values with a reduction ratio ε of 1.0 to 0.1. When the reduction ratio ε is not described in FIG. 4 (for example, ε = 0.95), the filter coefficient to be applied can be calculated by linear interpolation.

In the present embodiment, band limitation by an interpolation filter is executed using a one-dimensional filter for vertical and horizontal separation. Hereinafter, the band limitation in the horizontal direction will be described. The RGB gradation value of the pixel N to be subjected to the interpolation filter is (R _(N) , G _(N) , B _(N) ), and the RGB gradation value of the pixel (N−1) on the left side of the pixel N is (R ₍ N _N-1) , G _(N-1) , B _(N-1) ), and the RGB gradation values of the pixel (N + 1) on the right side of the pixel N are expressed as (R _{(N + 1)} , G _{(N + 1)} , B _{(N + 1)} ), and the RGB gradation values of the pixel (N + 2) on the right side of the pixel N + 1 are (R _{(N + 2)} , G _{(N + 2)} , B _{(N + 2)} ). In this case, the RGB gradation values (R ′ _(N) , G ′ _(N) , B ′ _(N) ) of the pixel N after band limitation by the interpolation filter are expressed by the following equations (1) to (3 ).
R ' _(N) = h _[-1] x R _(N-1) + h _[0] x R _(N) + h _[1] x R _{(N + 1)} + h _[2] x R _{(N + 2)} … (1)
G ' _(N) = h _[-1] x G _(N-1) + h _[0] x G _(N) + h _[1] x G _{(N + 1)} + h _[2] x G _{(N + 2)} … (2)
B ' _(N) = h _[-1] x B _(N-1) + h _[0] x B _(N) + h _[1] x B _{(N + 1)} + h _[2] x B _{(N + 2)} … (3)

  The band limitation in the vertical direction can be calculated similarly. Note that band limitation in either the vertical direction or the horizontal direction may be executed first.

  FIG. 5 is an explanatory diagram showing the filter characteristics of the interpolation filter having the filter coefficient of FIG. When the reduction rate ε is a small value, the graphs are crowded and difficult to see. Therefore, FIG. 5 shows only the reduction rate ε from 1.0 to 0.5. Since the number of filter coefficients (also called “the number of taps”) is finite, it is difficult to realize an ideal interpolation filter as shown in FIG. The filter characteristic is a curve in which the gain decreases on the high normalized frequency side, but the gain when the normalized frequency is ½ or more of the reduction ratio cannot be zero. In such a case, moire may be noticeable in a region where the normalized frequency is 1/2 or more of the reduction ratio.

FIG. 6 is an explanatory diagram showing an example of a filter coefficient used when band limitation is more strongly applied than that shown in FIG. Compared with the filter coefficient shown in FIG. 4, the value of h _[0] is small, and the values of h _[-1] and h _[+1] are large. Comparing the filter coefficient shown in FIG. 4 with the filter coefficient shown in FIG. 6, the filter coefficient shown in FIG. 6 has a smaller specific gravity of the pixel N in the weighted average than the filter coefficient shown in FIG. N-1) and the specific gravity of the pixel (N + 1) are large. As a result, when the interpolation filter having the filter coefficient shown in FIG. 6 is applied, the RGB gradation values of the pixels after the filter processing are averaged, compared to applying the interpolation filter having the filter coefficient shown in FIG. It can be said that the blur of the image becomes large.

  FIG. 7 is an explanatory diagram showing the filter characteristics of the interpolation filter having the filter coefficient of FIG. When the reduction rate ε is a small value, the graphs are crowded and difficult to see, and therefore, the reduction rate ε is shown from 1.0 to 0.5 as in FIG. Compared with the filter characteristic shown in FIG. 5, the filter characteristic shown in FIG. 7 has a smaller gain when the normalized frequency is ½ or more of the reduction ratio ε, and it can be said that moire is less noticeable. However, when compared with the filter characteristic shown in FIG. 5, the filter characteristic shown in FIG. 7 has a smaller gain even in a region where the normalized frequency is less than 1/2 of the reduction ratio ε, and the blur is increased. I can say that.

  FIG. 8 is an explanatory diagram illustrating an example of frequency characteristics of an image input to the projector. As can be seen from FIG. 8, the intensity (frequency) of the higher frequency component at the normalized frequency is smaller. Therefore, when the amount of deformation of the trapezoid correction that requires only a small amount of trapezoid correction from the uncorrected state is small (when the reduction ratio ε is close to 1), the image includes many frequency components to be cut off by the interpolation filter. Moire is not generated. In such a case, it is preferable to give priority to maintaining the sharpness of the image by relaxing the band limitation by the interpolation filter. Specifically, it is preferable to use the interpolation filter having the filter coefficient shown in FIG. 4 rather than the interpolation filter having the filter coefficient shown in FIG. On the contrary, when the amount of deformation of the trapezoidal correction is large, the frequency component to be cut off by the interpolation filter is included in the image, and moire may occur. Therefore, the interpolation filter having the filter coefficient shown in FIG. Preferably, the band is limited using an interpolation filter having filter coefficients shown in FIG.

  FIG. 9 is a block diagram of the projector 10 in the present embodiment. The projector 10 includes an angle information acquisition unit 100, a keystone correction parameter generation unit 110, a keystone correction unit 120, an image input unit 140, an image output unit 150, a liquid crystal panel (light modulation device) 160, and a projection optical system. 170, a light source 175, and a camera 180.

  The angle information acquisition unit 100 captures the test pattern projected on the screen 20 with the camera 180, and acquires angle information necessary for keystone correction based on the captured image. The angle information may be input by the user. In addition, the projector 10 includes a keystone correction knob or button, and when the user performs keystone correction by operating the keystone correction knob or button, angle information is obtained from the operation input to the knob or button. The unit 100 may be configured to obtain angle information.

  The keystone correction parameter generation unit 110 generates a perspective transformation coefficient for keystone correction. The trapezoidal correction parameter generation unit 110 includes a perspective transformation coefficient calculation unit 112 and a filter coefficient calculation unit 114. The perspective conversion coefficient calculation unit 112 acquires angle information from the angle information acquisition unit 100 and calculates a perspective conversion coefficient for keystone correction. Note that the perspective conversion coefficient may be determined based on the optical parameters of the projector and the projection angle. The filter coefficient calculation unit 114 calculates a filter coefficient for each reduction ratio ε in the keystone correction. The trapezoidal correction parameter generation unit 110 only needs to operate when the angle information changes, and may be implemented by software.

  The trapezoid correction unit 120 performs keystone correction with reference to the perspective transformation coefficient and the filter coefficient. The trapezoid correction unit 120 includes a perspective conversion coefficient register 122, a filter coefficient table 124, a reduction ratio calculation unit 126, a frame memory 128, and a calculation unit 130. Since the keystone correction unit 120 performs keystone correction for each frame, the keystone correction unit 120 may be implemented by, for example, an ASIC (Application Specific Integrated Circuit).

The perspective conversion coefficient register 122 stores the perspective conversion coefficient calculated by the perspective conversion coefficient calculation unit 112. The filter coefficient table 124 is a table that stores filter coefficients for each reduction rate ε. Reduction rate calculator 126 calculates the reduction ratio epsilon at the pixel of a plurality of predetermined positions on the basis of the perspective transformation coefficients, the minimum value epsilon _min reduction ratio epsilon (hereinafter, the minimum value epsilon _min reduction factor epsilon ' Minimum reduction rate ε _min "). In the trapezoidal correction, the reduction rate ε is minimized at one of the four vertices of the quadrangle. At one of the other three points, the reduction ratio ε is maximum, and the values of the reduction ratio ε at the other two points and the points inside the rectangle are between the maximum and minimum values. Therefore, the reduction rate calculation unit 126 may calculate the reduction rate ε of these four vertices with a predetermined position as four vertices of a rectangle.

  FIG. 10 is an explanatory diagram showing how to obtain the reduction ratio ε. 10A shows the first image Pic0 on the liquid crystal panel 160 before keystone correction, and FIG. 10B shows the second image Pic0 'on the liquid crystal panel 160 after keystone correction. Assuming that the four vertices of the liquid crystal panel 160 are points P0, Q0, R0, and S0, the first image Pic0 on the liquid crystal panel 160 before keystone correction is shown in FIG. It is a rectangle having R0 and S0 as vertices. When the first image Pic0 is projected onto the screen 20, the image Pic1 on the screen 20 becomes a distorted square due to the positional relationship between the projector 10 and the screen 20, as shown in FIG. The trapezoid correction unit 120 performs a trapezoid correction process on the first image Pic0, and a second image Pic0 ′ having four points P0 ′, Q0 ′, R0 ′, S0 ′ on the liquid crystal panel 160 as vertices. Form. Points P0 ', Q0', R0 ', S0' are points obtained by trapezoidally correcting points P0, Q0, R0, S0, respectively. The second image Pic0 'after the keystone correction is an image smaller than the first image Pic0 before the keystone correction. When the second image Pic <b> 0 ′ is projected on the screen 20, a rectangular image Pic <b> 2 (FIG. 1) is formed on the screen 20. In this way, the trapezoidal correction unit 120 generates a second image Pic0 'obtained by deforming the first image Pic0.

The reduction ratio calculation unit 126 calculates a reduction ratio ε for the perspective transformation from the point P0 to the point P0 ′. The reduction ratio calculation unit 126 sets the eight pixels around the point P0 ′ of the second image Pic0 ′ after the perspective transformation as points P1 ′ to P8 ′ in the clockwise order from the upper left. Next, the reduction ratio calculation unit 126 performs reverse perspective transformation on the points P1 ′ to P8 ′, and calculates the coordinates of the points P1 to P8 of the first image Pic0. The points P1 to P8 and the points P1 ′ to P8 ′ have a relationship in which the points P1 to P8 are converted into points P1 ′ to P8 ′ by perspective transformation, respectively. The reduction ratio calculation unit 126 includes a distance x1 between the points P4 and P8, a distance y1 between the points P2 and P6, a distance x2 between the points P4 ′ and P8 ′, and a point P2. The distance y2 between “and the point P6” is calculated. The reduction rate calculation unit 126 calculates the horizontal reduction rate εPx and the vertical reduction rate εPy by the following equations (4) and (5).
Horizontal direction: εPx = x2 / x1 (4)
Longitudinal direction: εPy = y2 / y1 (5)
Since x2 ≦ x1 and y2 ≦ y1, the reduction ratios εPx and εPy are values of 1 or less. The reduction ratio calculation unit 126 also performs horizontal and vertical reduction for perspective transformation from the point Q0 to the point Q0 ′, perspective transformation from the point R0 to the point R0 ′, and perspective transformation from the point S0 to the point S0 ′. The rate ε is calculated similarly. The minimum reduction ratio ε _min is obtained separately in the horizontal direction and the vertical direction.

  The image input unit 140 in FIG. 9 receives an input of an image projected by the projector 10. The frame memory 128 is a memory that stores an input image. The calculation unit 130 performs perspective transformation on the image in the frame memory and applies an interpolation filter. In the present embodiment, the interpolation filter is a one-dimensional filter that separates the x direction (horizontal direction) and the y direction (vertical direction). The arithmetic unit 130 first performs perspective transformation, and then reads the filter coefficient in the x direction from the filter coefficient table 124 and applies an interpolation filter. Next, similarly in the y direction, the filter coefficient is read from the filter coefficient table 124 and an interpolation filter is applied. In the filter coefficient table 124, for example, a filter coefficient is defined for each reduction ratio of 0.1. If a corresponding reduction ratio is not defined, interpolation may be performed using the upper and lower filter coefficients. Good.

  The image output unit 150 outputs the second image Pic <b> 0 ′ subjected to the perspective transformation and the filtering process to the liquid crystal panel 160. The output second image Pic0 'is a distorted rectangle as shown in FIG. The light source 175 emits light to the liquid crystal panel 160. The liquid crystal panel 160 modulates the light emitted from the light source 175 according to the image signal, and emits light representing an image corresponding to the image signal. The projection optical system 170 is an optical system that projects light emitted from the liquid crystal panel 160 onto a screen. As a result, the second image Pic0 'formed on the liquid crystal panel 160 is projected onto the screen 20 as an image Pic1'. The light source 175, the liquid crystal panel 160, and the projection optical system 170 constitute a projection unit that projects an image.

FIG. 11 is a flowchart of filter coefficient calculation. In step S100, the reduction ratio calculation unit 126 calculates the minimum image reduction ratio ε _min in the perspective transformation. As described above, in the perspective transformation, the reduction rate ε is minimized at any of the four vertices of the image. Therefore, the reduction rate calculation unit 126 determines the reduction rates ε of the four vertices P0, Q0, R0, and S0. The minimum reduction rate ε _min may be obtained by calculation. The minimum reduction rate ε _min is determined for each of the horizontal direction and the vertical direction.

  In step S110, the calculation unit 130 calculates a filter coefficient used for the filter processing of each pixel. The cutoff frequency theoretically determined corresponding to the reduction ratio ε of each pixel is Fc, and the cutoff frequency used for the filter processing is Fc ′. The theoretical cutoff frequency is ½ of the reduction ratio ε as described in FIG. In the present embodiment, the calculation unit 130 uses the filter coefficient of the adjusted cutoff frequency Fc ′ calculated by Fc ′ = Fc × α. Here, α is a cutoff frequency adjustment coefficient, and is determined from the graph shown in FIG.

FIG. 12 is an explanatory diagram showing the relationship between the minimum reduction rate ε _min and the cutoff frequency adjustment coefficient α. The horizontal axis is the minimum reduction ratio ε _min , and the vertical axis is the cut-off frequency adjustment coefficient α. When the minimum reduction ratio ε _min is less than a predetermined threshold value S, the cutoff frequency adjustment coefficient α is 0.5. When the minimum reduction rate ε _min is equal to or greater than the threshold value S, the cutoff frequency adjustment coefficient α when the minimum reduction rate ε _min is the threshold value S is 0.5, and the cut is performed in proportion to the value of the minimum reduction rate ε _min. The off-frequency adjustment coefficient α is set to increase. When the minimum reduction ratio ε _min is 1.0, the cutoff frequency adjustment coefficient α is 1.0.

The relationship between the minimum reduction ratio ε _min and the cutoff frequency Fc ′ can be expressed by the following equations (6) and (7).
Minimum reduction ratio ε _min is less than threshold S: Fc ′ = (ε / 2) × 0.5 (6)
Minimum reduction ratio ε _min is equal to or greater than threshold S: Fc ′ = (ε / 2) × α (7)

As can be seen from the above equations (6) and (7), when the minimum reduction ratio ε _min is equal to or larger than the threshold value S, that is, when the image is not reduced much in the trapezoidal correction, the cutoff frequency Fc ′ is expressed by the equation (7). Since it is calculated, it is larger than the cutoff frequency Fc ′ calculated by the equation (6). As a result, the interpolation filter is weakly applied to the image. Specifically, when the minimum reduction rate ε _min is equal to or greater than the threshold S, the calculation unit 130 acquires the filter coefficient from the filter coefficient table 124 with the reduction rate as (ε × α), and the minimum reduction rate ε _min is the threshold value. If it is less than S, the filter coefficient is obtained from the filter coefficient table 124 with the reduction ratio (ε × 0.5). The calculation unit 130 performs filter processing using the filter coefficient.

As described above, according to the present embodiment, when the minimum reduction ratio ε _min is equal to or larger than the threshold S, that is, when the image is not reduced much in the trapezoidal correction, the cutoff frequency adjustment coefficient is larger than that of the comparative example, and the filter processing is weak. Done. That is, as shown in FIG. 8, since the intensity (frequency) of the high frequency component at the normalized frequency is smaller, the amount of deformation of the trapezoid correction that requires only a slight trapezoid correction from the uncorrected state is small (the reduction ratio ε is When the frequency is close to 1), the frequency component to be cut off by the interpolation filter is not included in the image, and moire is unlikely to occur. Therefore, when the image is not reduced very much in the trapezoidal correction, it is possible to relax the band limitation by the interpolation filter and maintain the sharpness of the image.

In the above embodiment, the arithmetic unit 130, when the minimum reduction ratio epsilon _min is less than the threshold value S of the formula (6), when the minimum reduction ratio epsilon _min is equal to or larger than the threshold S has been using the equation (7) The equation (7) may be applied to all the reduction ratios ε without providing the threshold S. In this case, the cut-off frequency adjustment coefficient α may be a larger value as the minimum reduction ratio ε _min is larger. Further, the minimum reduction ratio ε _min and the cutoff frequency adjustment coefficient α do not necessarily have a linear relationship, and for example, a quadratic curve may be formed. As described above, even when the threshold value S is not provided, the filter coefficient is selected so that the band limitation of the interpolation filter is weakened as the minimum reduction ratio ε _min is large. In addition, moire can be suppressed.

  The value of the threshold S is different from a plurality of projection modes of the projector 10, for example, a projection mode according to the content of an image to be projected such as landscape or presentation, or a projection mode such as a still image or a moving image. It may be set. The frequency characteristics of the image shown in FIG. 8 vary depending on the content of the image. For example, when the frequency characteristic of the image is acquired by Fourier transform and the intensity (frequency) of the high frequency component is small, the value of the threshold S may be lowered as compared with the case where the intensity (frequency) of the high frequency component is large. In other words, the threshold value S may be set to increase as the strength (frequency) of the high-frequency component decreases according to the strength (frequency) of the high-frequency component.

In this embodiment, the horizontal direction and the vertical direction are separated and the one-dimensional interpolation filter is applied based on the respective reduction ratios. However, the horizontal direction and the vertical direction are not separated, and a two-dimensional interpolation filter is used. Also good. In this case, as the minimum reduction ratio ε _min , the smallest value of the reduction ratios in the horizontal direction and the reduction ratio in the vertical direction of each of the four vertices may be used.

  In the present embodiment, the case where the number of taps is four has been described as an example for the one-dimensional interpolation filter, but the number of taps may be three, for example. In this case, the RGB gradation value of the pixel N after band limitation is a weighted average of the RGB gradation values of the three pixels (N−1), N, and (N + 1). Further, the number of taps may be five or more. When the number of taps is 5, the RGB gradation value of the pixel N after band limitation is applied to the five pixels of the pixels (N−2), (N−1), N, (N + 1), and (N + 2). This is a weighted average of RGB gradation values.

  The embodiments of the present invention have been described above based on some examples. However, the above-described embodiments of the present invention are for facilitating the understanding of the present invention and limit the present invention. It is not a thing. The present invention can be changed and improved without departing from the spirit and scope of the claims, and it is needless to say that the present invention includes equivalents thereof.

  The configuration of the light modulation device of the projector is not limited to the configuration described in the above embodiment. In the present embodiment, the light modulation device has been described as the transmissive liquid crystal panel 160. However, when a liquid crystal panel is used as the light modulation device, the light modulation device may be a transmissive liquid crystal panel as in the present embodiment. A reflective liquid crystal panel may also be used. In the case of the reflection type, the light modulation device may be a digital mirror device (DMD) having a plurality of micromirrors arranged in an array instead of the liquid crystal panel. Furthermore, the light modulation device may use a self-luminous element such as an organic EL. In this case, a light source for illuminating the light modulation device is not necessary.

  As the light source, a light source composed of a lamp such as a xenon lamp or an ultrahigh pressure mercury lamp, or a solid light source such as an LED (Light Emitting Diode) or a laser light source can be employed. In this embodiment, the screen 20 is used as the projection surface, but a wall may be used.

DESCRIPTION OF SYMBOLS 10 ... Projector 20 ... Screen 100 ... Angle information acquisition part 110 ... Trapezoid correction parameter generation part 112 ... Perspective conversion coefficient calculation part 114 ... Filter coefficient calculation part 120 ... Trapezoid correction part 122 ... Perspective conversion coefficient register 124 ... Filter coefficient table 126 ... Reduction ratio calculation unit 128 ... frame memory 130 ... calculation unit 140 ... image input unit 150 ... image output unit 160 ... liquid crystal panel (light modulation device)
170 ... Projection optical system 175 ... Light source 180 ... Camera

Claims (6)

A projector,
A trapezoidal correction unit that generates a second image obtained by deforming the first image by an interpolation process using an interpolation filter in order to perform a trapezoidal correction of the image projected on the projection surface according to the first image; ,
A light modulation device that emits light representing the second image;
A projection optical system that projects light from the light modulation device onto the projection surface;
With
The trapezoidal correction unit is
Calculating a reduction ratio when transforming from the first image to the second image at a plurality of predetermined points in the first image;
The interpolation filter is adjusted to weaken the band limitation of the interpolation filter as the minimum value of the reduction ratio is larger ,
When the minimum value of the reduction ratio is equal to or greater than a predetermined threshold, the adjustment coefficient represented by the ratio of the cutoff frequency of the interpolation filter to the reduction ratio is increased as the minimum value of the reduction ratio is larger. To reduce the bandwidth limit,
When the minimum value of the reduction ratio is less than the threshold, the adjustment coefficient is set to a predetermined value .
projector. The projector according to claim 1.
The projector, wherein the predetermined plurality of points are four vertices of the first image. The projector according to claim 1 or 2 ,
The projector, wherein the threshold value is set to a different value for a plurality of projection modes of the projector. An image processing method in a projector,
(A) A step of generating a second image obtained by deforming the first image by interpolation processing using an interpolation filter in order to perform trapezoidal correction of the image projected on the projection surface according to the first image. When,
(B) emitting light representing the second image from the light modulation device;
(C) projecting light from the light modulation device onto the projection surface;
With
The step (a)
(A-1) calculating a reduction ratio when the first image is transformed into the second image at a plurality of predetermined points in the first image;
(A-2) adjusting the interpolation filter so that the band limitation of the interpolation filter is weakened as the minimum value of the reduction ratio is larger;
Equipped with a,
In the step (a-2),
When the minimum value of the reduction ratio is equal to or greater than a predetermined threshold, the adjustment coefficient represented by the ratio of the cutoff frequency of the interpolation filter to the reduction ratio increases as the minimum value of the reduction ratio increases. The bandwidth limit is weakened,
Wherein when the minimum reduction ratio is smaller than the threshold value, the adjustment factor Ru is a predetermined value, the image processing method. The image processing method according to claim 4 ,
The image processing method, wherein the predetermined plurality of points are four vertices of the first image. The image processing method according to claim 4 or 5 ,
The image processing method, wherein the threshold value is set to a different value for a plurality of projection modes of the projector.

Images