Projector automatic trapezoid correction method and system and medium

Abstract

The invention provides an automatic trapezoid correction method and system for a projector and a medium. The method comprises the following steps: acquiring attitude information of the projector and spatial information of the projector and a projection surface; under the condition that optical emission parameters of the projector are known, a projection geometric model is established based on the attitude information and the space information, and transformation parameters used for projector trapezoidal correction are generated accordingly. And controlling the projector to execute a trapezoidal correction action based on the transformation parameter. And controlling the projector to project a preset calibration pattern, acquiring an image of the calibration pattern by adopting the image acquisition device after the trapezoidal correction action is executed, extracting image features, and acquiring geometric parameters corresponding to the actual projection picture based on the image features. And determining a calibration error based on the difference between the geometric parameter and the transformation parameter, and generating a compensation parameter according to the calibration error. And writing the compensation parameter into the projector so as to compensate the transformation parameter when the projector performs trapezoidal correction. Through the steps, the projector generates a corresponding correction coefficient.

Description

Technical Field

[0001] This application relates to the field of projection display, and more particularly to an automatic keystone correction method, system, and medium for a projector. Background Technology

[0002] Automatic keystone correction is a key function of modern projection equipment, designed to compensate for geometric distortion (trapezoidal distortion) caused by the projection angle when the projector is not perpendicular to the projection surface, through digital image transformation. For ultra-short-throw projectors, their special optical design projects large images over extremely short distances, making their optical paths and spatial geometry more complex than those of conventional projectors.

[0003] Currently, the most common solution for automatic keystone correction in projectors is the lookup table method based on preset parameter tables. This method involves measuring a few prototypes during the R&D phase to create a lookup table of posture, distance, and correction parameters, which is then fixed in all products of the same model. However, because it cannot cover the unique hardware tolerances of each piece of equipment on the production line, the lookup table method provides an "average" solution. Its correction accuracy is random for individual devices, resulting in unstable effects and a poor user experience. Because none of the above methods can effectively address individual production differences, many ultra-short-throw projector products either abandon automatic keystone correction altogether or maintain its effect at a low level, severely restricting the improvement of product usability and user experience.

[0004] Therefore, an automatic keystone correction method for projectors is needed to generate corresponding correction coefficients. Summary of the Invention

[0005] In view of this, it is necessary to provide an automatic keystone correction method, system, and medium for projectors to solve the above problems.

[0006] Embodiments of this application provide an automatic keystone correction method, system, and medium for a projector, applicable to ultra-short-throw projectors of the same model in production. The method includes the following steps: Obtain the attitude information of the projector and the spatial information of the projector and the projection surface; Given the optical emission parameters of the projector, a projection geometry model is established based on the attitude information and the spatial information, and transformation parameters for trapezoidal correction of the projector are generated accordingly. Based on the transformation parameters, the projector is controlled to perform a trapezoidal correction action; The projector is controlled to project a preset calibration pattern. After the trapezoidal correction action is performed, an image acquisition device is used to acquire an image of the calibration pattern and extract image features. Based on the image features, geometric parameters corresponding to the actual projected image are obtained. The calibration error is determined based on the difference between the geometric parameters and the transformation parameters, and compensation parameters are generated based on the calibration error. The compensation parameters are written into the projector to compensate for the transformation parameters when the projector performs keystone correction.

[0007] In at least one embodiment of this application, obtaining the attitude information includes the following steps: Place the projector on a horizontal reference plane; Acquire the triaxial acceleration sampling sequence of the inertial measurement unit within a preset sampling time; The average value of the three-axis acceleration sampling sequence is calculated to obtain the three-axis acceleration bias. In subsequent attitude calculations, the corrected acceleration value is obtained by subtracting the triaxial acceleration offset from the real-time triaxial acceleration value, and the roll and pitch angles are calculated based on the corrected acceleration value.

[0008] In at least one embodiment of this application, obtaining spatial information includes the following steps: Acquire dot matrix ranging data output by the ranging sensor; The dot matrix ranging data is converted into a spatial point set based on the field of view parameters of the ranging sensor; Points that exceed a preset distance range or whose distance measurement is invalid are removed from the spatial point set; Fit the projection plane to the remaining point set to obtain the plane parameters.

[0009] In at least one embodiment of this application, the fitting of the projection plane includes the step of: Select no less than a preset number of valid points from the set of spatial points; A robust plane fitting strategy is used to screen out outliers and obtain the plane normal vector; The plane normal vector is transformed into a coordinate system established by the inertial measurement unit, and the yaw angle is calculated based on the plane normal vector and the horizontal reference direction of the coordinate system.

[0010] In at least one embodiment of this application, the step of establishing the projection geometric model and generating transformation parameters includes the following steps: Based on the optical emission parameters, multiple projection boundary rays corresponding to the boundary of the display panel are determined; Calculate the intersection point of the projection boundary ray and the projection plane; The intersection points are determined as the set of boundary feature points of the projection area, and the target correction area is determined accordingly.

[0011] In at least one embodiment of this application, determining the target correction area includes the step of: The trapezoidal boundary is determined by the boundary feature points of the projection area; Find the inscribed rectangle with the largest area within the trapezoidal boundary, and use the vertices of the inscribed rectangle as feature points of the target correction region.

[0012] In at least one embodiment of this application, generating transformation parameters includes the step of: Establish the correspondence between the reference feature points in the display panel coordinate system and the feature points of the target correction area; Solve for the homography transformation matrix based on the aforementioned correspondence; The geometric transformation parameters of the display panel image are output based on the homography transformation matrix and used as control parameters for the trapezoidal correction action.

[0013] In at least one embodiment of this application, generating compensation parameters based on calibration error includes the following steps: The calibration error is obtained and the pitch angle compensation value is calculated under at least one positive projection attitude; The calibration error was obtained under multiple different rotation angle attitudes, and the corresponding yaw angle compensation was calculated. The rotation angle and the yaw angle compensation amount are constructed into a fitting dataset, and the compensation function coefficients are solved for the left turn and right turn conditions respectively. During production line calibration, the pitch angle compensation value is adjusted based on the calibration error of the projector to be calibrated, and the constant term of the left turn compensation function or right turn compensation function is corrected and then written into the projector.

[0014] An automatic keystone correction system for a projector, applied to any of the automatic keystone correction methods for projectors described above.

[0015] An automatic keystone correction medium for a projector, applied to any of the above-described automatic keystone correction methods for projectors. The aforementioned automatic keystone correction method for projectors does not consider the process complete after the system and medium undergo preliminary correction using a geometric model. Instead, it actively projects a known calibration pattern and acquires the actual geometric parameters of the corrected image through image acquisition. Subsequently, the solution converts this measured error into compensation parameters specific to that particular device and permanently writes them into its memory. Therefore, whenever the user performs keystone correction in any future usage scenario, the system will call upon these individual compensation parameters to correct the general model calculation results in real time. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the automatic keystone correction method for projectors described in this application. Detailed Implementation

[0017] The embodiments of this application will now be described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0018] It should be noted that when a component is considered to be "connected" to another component, it can be directly connected to the other component or may also have an intervening component. When a component is considered to be "placed" on another component, it can be directly placed on the other component or may also have an intervening component. The terms "top," "bottom," "upper," "lower," "left," "right," "front," "back," and similar expressions used in this article are for illustrative purposes only.

[0019] Embodiments of this application provide an automatic keystone correction method, system, and medium for a projector, applicable to ultra-short-throw projectors of the same model in production. The method includes the following steps: S10: Obtain the attitude information of the projector and the spatial information of the projector and the projection surface; S20: Given the optical emission parameters of the projector, establish a projection geometry model based on the attitude information and the spatial information, and generate transformation parameters for trapezoidal correction of the projector accordingly. S30: Control the projector to perform a trapezoidal correction action based on the transformation parameters; S40: Control the projector to project a preset calibration pattern. After the trapezoidal correction action is performed, use an image acquisition device to acquire an image of the calibration pattern and extract image features. Based on the image features, obtain the geometric parameters corresponding to the actual projected image. S50: Determine the calibration error based on the difference between the geometric parameters and the transformation parameters, and generate compensation parameters based on the calibration error; S60: Write the compensation parameters into the projector to compensate the transformation parameters when the projector performs keystone correction.

[0020] The aforementioned automatic keystone correction method for projectors does not consider the process complete after the system and medium undergo preliminary correction using a geometric model. Instead, it actively projects a known calibration pattern and acquires the actual geometric parameters of the corrected image through image acquisition. Subsequently, the solution converts this measured error into compensation parameters specific to that particular device and permanently writes them into its memory. Therefore, whenever the user performs keystone correction in any future usage scenario, the system will call upon these individual compensation parameters to correct the general model calculation results in real time.

[0021] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0022] Please see Figure 1 This application provides an automatic keystone correction method for a projector, applicable to ultra-short-throw projectors of the same model in production. The method includes the following steps: S10: Obtain the attitude information of the projector and the spatial information of the projector and the projection surface.

[0023] In this step, it should be noted that the attitude information is used to characterize the current spatial attitude state of the projector, preferably including attitude quantities such as roll angle and pitch angle, which are used to describe the degree of tilt of the projector relative to the direction of gravity.

[0024] The spatial information of the projector and the projection surface is used to characterize the geometric relationship between them, preferably including the position parameters of the projection surface, the normal vector of the projection surface, or the orientation relationship of the projector relative to the projection surface. By first acquiring attitude and spatial information, subsequent trapezoidal correction can be based on quantifiable physical quantities, making the correction process independent of manual experience or averaged lookup tables. This provides a consistent input benchmark for mass-produced ultra-short-throw projectors of the same model, reduces the risk of model drift caused by single-unit assembly tolerances, and improves the repeatability and stability of subsequent transformation parameter calculations.

[0025] S20: Given the optical emission parameters of the projector, establish a projection geometry model based on the attitude information and the spatial information, and generate transformation parameters for trapezoidal correction of the projector accordingly.

[0026] In this step, it should be noted that the optical emission parameters are used to describe the emission geometry of the optical engine, such as the direction of light emission, the nominal mapping relationship between the display panel and the projection optical path, and the outgoing vector corresponding to the projection boundary light.

[0027] The purpose of establishing a projection geometry model based on attitude and spatial information is to convert the "geometric deformation of the image on the projection surface caused by the tilt of the projector" into a computable geometric relationship, thereby generating transformation parameters for trapezoidal correction.

[0028] The transformation parameters are preferably parameters that perform geometric transformations on the display panel image, such as homography transformation matrices or equivalent coordinate mapping parameters.

[0029] This step allows for the generation of a set of initial correction parameters with clear physical meaning from the geometric model without relying on the camera to directly infer all correction values. This ensures that subsequent calibration errors mainly reflect system deviations such as single-machine hardware tolerances and attitude measurement residual errors, thereby improving the convergence and stability of the overall calibration.

[0030] S30: Control the projector to perform a trapezoidal correction action based on the transformation parameters.

[0031] In this step, it should be noted that the key is to first perform a trapezoidal correction based on the model calculation results, so that the projected image enters the correction state predicted by the model.

[0032] Compared with the traditional method of directly acquiring calibration patterns and calculating errors on uncorrected images, performing a trapezoidal correction first can significantly reduce the degree of distortion of the projected image, making the shape of the calibration pattern on the projection surface closer to the ideal state, thereby improving the accuracy and robustness of image feature extraction. For example, it can reduce false detections and false negatives caused by severe distortion, boundary clipping, or local stretching in checkerboard corner detection.

[0033] At the same time, this step can also transform the subsequent error calculation from full correction of large distortion state to fine compensation of model residuals, thereby improving the stability of compensation parameter fitting and better meeting the cycle time and consistency requirements of mass production calibration.

[0034] S40: Control the projector to project a preset calibration pattern. After the trapezoidal correction action is performed, use an image acquisition device to acquire an image of the calibration pattern and extract image features. Based on the image features, obtain the geometric parameters corresponding to the actual projected image.

[0035] In this step, it should be noted that the preset calibration pattern is preferably a checkerboard, concentric circle array, or other pattern that facilitates the extraction of geometric features. Among them, the checkerboard pattern is suitable for obtaining stable geometric parameters due to its large number of corner points and regular structure. The image acquisition device is preferably a production line camera or an external camera. The acquired calibration pattern image is processed to extract image features, which preferably include corner coordinates, boundary lines, grid spacing, orthogonality error, etc.

[0036] The purpose of obtaining the geometric parameters corresponding to the actual projected image based on image features is to obtain a quantitative description of the "real image geometry," providing a true measurement benchmark for subsequent calibration error calculation. Since a trapezoidal correction operation has already been performed in the previous step, the actual acquired calibration pattern is closer to the ideal shape, and the corner point positioning accuracy is higher, thus making the geometric parameters more accurate. This improves the reliability of the calibration error calculation and reduces the impact of noise on the solution of compensation parameters.

[0037] S50: Determine the calibration error based on the difference between the geometric parameters and the transformation parameters, and generate compensation parameters based on the calibration error.

[0038] In this step, it should be noted that the calibration error is used to characterize the deviation between the model's predicted correction result and the actual corrected geometric shape of the real image. Its sources usually include single-machine assembly tolerance, optomechanical installation deviation, sensor zero-bias residual error, and the difference between the nominal value of optical parameters and the actual machine.

[0039] By comparing the actual geometric parameters with the transformed parameters, the deviation caused by the combination of the above complex factors can be attributed to a calculable calibration error.

[0040] Furthermore, generating compensation parameters based on calibration errors converts the calibration errors into storable and reusable correction values, enabling each projector in mass production of the same model to obtain its own compensation parameters, rather than using uniform lookup table data.

[0041] The beneficial effect of this step is that it can solidify random individual differences into callable compensation parameters, thereby significantly improving the consistency of subsequent trapezoidal correction and avoiding the problem of large differences in correction effects between different units of the same model in user scenarios.

[0042] S60: Write the compensation parameters into the projector to compensate the transformation parameters when the projector performs keystone correction.

[0043] In this step, it's important to note that after the compensation parameters are written into the projector's memory, the projector can use these parameters for secondary correction based on the transformation parameters generated by the general geometric model when any subsequent keystone correction is triggered. This offsets system deviations caused by individual hardware tolerances. Unlike temporary adjustments only at the production end, writing the compensation parameters permanently solidifies the individual projector's characteristics, ensuring stable keystone correction even when users change the projection position or fine-tune the placement angle. The benefits of this step include significantly improving mass production consistency and user experience for ultra-short-throw projectors of the same model, reducing rework and debugging costs, and providing a data foundation for subsequent quality traceability.

[0044] The process of obtaining attitude information includes the following steps: S11: Place the projector on a horizontal reference plane.

[0045] In this step, it should be noted that the horizontal reference plane is used to provide a unified attitude "zero-point reference" so that the direction of gravity sensed by the inertial measurement unit in a static state can form a repeatable correspondence with the structural reference direction of the projector.

[0046] By placing the projector on a horizontal reference plane, the impact of tilting, uneven surfaces, or human placement errors on subsequent calibration can be minimized, thus providing a prerequisite for obtaining stable and reliable triaxial acceleration reference values. The benefit of this step is that it establishes a consistent measurement attitude reference for subsequently calculating triaxial acceleration offsets, ensuring that mass-produced equipment of the same model has a consistent calibration starting point during production line calibration, and improving the repeatability of calibration results.

[0047] S12: Acquire the triaxial acceleration sampling sequence of the inertial measurement unit within the preset sampling time.

[0048] In this step, it should be noted that the triaxial acceleration values ​​output by the inertial measurement unit may fluctuate in a short period of time due to sensor noise, micro-vibration, and quantization error. Therefore, obtaining the sampling sequence by continuously sampling within a preset sampling time is beneficial for statistically suppressing random noise.

[0049] The preset sampling duration can be set according to the production line cycle time and noise level, such as a time window of tens of milliseconds to several seconds.

[0050] This step obtains acceleration information by sampling a sequence rather than a single sample value, making the bias value calculated in subsequent calculations more stable. This reduces the angle jitter caused by noise in attitude calculation and provides a more reliable attitude input for the trapezoidal correction model.

[0051] S13: Calculate the mean of the three-axis acceleration sampling sequences to obtain the three-axis acceleration bias.

[0052] In this step, it should be noted that the so-called triaxial acceleration bias can be understood as the fixed deviation that still exists in the sensor under the theoretically horizontal static state. This deviation may come from the sensor zero bias, installation tilt angle, or structural assembly error.

[0053] By averaging the triaxial acceleration sampling sequences, the influence of random noise can be eliminated, and an offset closer to the fixed deviation can be extracted. The benefits of this step are: obtaining an offset reference that can be used for subsequent uniform subtraction, so that attitude calculation is no longer directly affected by the sensor's zero offset, thereby improving the accuracy and consistency of roll and pitch angle calculations, and laying the foundation for stable trapezoidal correction of "same model, different machines" in subsequent mass production.

[0054] S14: In subsequent attitude calculations, the corrected acceleration value is obtained by subtracting the triaxial acceleration offset from the real-time triaxial acceleration value, and the roll angle and pitch angle are calculated based on the corrected acceleration value.

[0055] In this step, it should be noted that the process of subtracting the offset from the real-time triaxial acceleration value is essentially a zero-point calibration of the inertial measurement unit output, so that the corrected acceleration value is closer to the real gravity component, thereby improving the accuracy of inverse calculation of attitude angle from acceleration component.

[0056] When calculating roll and pitch angles based on the corrected acceleration values, commonly used attitude calculation methods in this field can be employed, such as calculating roll and pitch angles using the component relationships of the gravity vector in the projector coordinate system. This step makes the calculated roll and pitch angles insensitive to sensor bias and assembly errors, thereby reducing the deviation of trapezoidal correction parameters caused by attitude errors, reducing the problem of slight tilt, non-parallelism of top and bottom edges, or corner offset still appearing after image correction, and improving the convergence and stability of subsequent compensation parameter fitting.

[0057] The process of obtaining spatial information includes the following steps: S15: Acquire the dot matrix ranging data output by the ranging sensor.

[0058] In this step, it is necessary to obtain the distance values ​​from multiple sampling directions simultaneously, for example, in the form of an m×n dot matrix. Compared with single-point ranging, dot matrix ranging can provide spatial morphology information of the projection surface within the field of view, providing sufficient data support for subsequent fitting of the projection surface plane.

[0059] This step quickly obtains distance information from multiple points covering a certain range without relying on external high-precision measuring equipment, making the geometric estimation of the projection surface more robust and reducing spatial information distortion caused by local occlusion or single-point errors. S16: Based on the field of view parameters of the ranging sensor, the dot matrix ranging data is converted into a spatial point set.

[0060] In this step, it should be noted that the dot matrix ranging data is essentially a distance value, and it needs to be further converted into three-dimensional spatial coordinate points for use in plane fitting. The so-called field of view parameters typically include: horizontal field of view angle, vertical field of view angle, angular distribution relationship of each sampling direction of the dot matrix, and calibration relationship between the sensor coordinate system and the projector structural coordinate system, etc.

[0061] By using the field of view parameters, the ranging direction corresponding to each lattice unit can be determined. Then, by combining the direction with the distance value, the three-dimensional coordinates of the spatial point set can be obtained.

[0062] The beneficial effect of this step is that it transforms the originally scattered distance data into a spatial point cloud under a unified coordinate system, thereby giving the subsequent plane fitting a clear geometric meaning and providing basic data for further calculations such as the projection plane normal vector and the orientation relationship between the projector and the projection plane.

[0063] S17: Remove points from the spatial point set that exceed the preset distance range or whose distance measurement is invalid.

[0064] In this step, it should be noted that the ranging sensor may be affected by factors such as the reflectivity of the projection surface material, ambient light, obstructions, and weak edge reflections in the actual production line environment. This can lead to invalid ranging at some points, such as empty, zero, saturated, or abnormally fluctuating values, or outliers that significantly deviate from the normal projection distance range. By setting a preset distance range and removing invalid points, outliers and erroneous points can be effectively filtered out, preventing them from causing serious deviations in the planar parameters after participating in planar fitting.

[0065] This step improves the quality of the spatial point set, making the plane fitting more stable and more resistant to interference, reducing projection surface estimation errors caused by a few outliers, thereby improving the reliability and consistency of subsequent trapezoidal correction transformation parameters.

[0066] S18: Fit the projection plane to the remaining point set to obtain the plane parameters.

[0067] In this step, it's important to note that the spatial point set after filtering out invalid points can represent the geometry of the projection surface within the ranging field of view quite well. By performing plane fitting on this point set, the plane parameters of the projection surface can be obtained, such as the plane normal vector and plane equation parameters. These plane parameters can not only characterize the position and orientation of the projection surface relative to the projector, but also be used to calculate the yaw angle of the projector relative to the projection surface, and together with attitude information, to establish the projection geometry model. The beneficial effects of this step are: obtaining spatial information of the projection surface through multi-point fitting is less susceptible to noise compared to single-point distance inference, and it can provide the best overall plane estimate even when the projection surface has slight unevenness or local errors, thereby improving the computational accuracy and robustness of automatic keystone correction.

[0068] The fitting of the projection plane includes the following steps: Select no fewer than a preset number of valid points from the set of spatial points.

[0069] In this step, it's important to note that even after removing invalid and out-of-range points, the spatial point set may still have insufficient points, overly concentrated point distribution, or missing points in certain areas. This can affect the stability and identifiability of the plane fitting. Therefore, this step sets a lower limit on the number of points to ensure sufficient data for fitting and avoids including obviously unreliable or duplicate points. The preset number can be set based on the sensor array size and field of view coverage; for example, it should include at least several non-collinearly distributed points. This step ensures that the fitting input point set has sufficient constraint information, making the obtained plane parameters more stable and preventing plane normal vector jitter due to insufficient points or distribution degradation, thereby reducing fluctuations in subsequent yaw angle calculations.

[0070] A robust plane fitting strategy is used to screen out outliers and obtain the plane normal vector.

[0071] In this step, it's important to note that although invalid and out-of-range points have been removed, a small number of outliers may still be present in the spatial point set due to factors such as differences in projection surface material reflection, ambient light interference, local obstructions, and sensor noise. These outliers can significantly skew the results of ordinary least squares plane fitting. A robust plane fitting strategy refers to a fitting method that is insensitive to outliers during the fitting process and can automatically identify and remove them. Examples include fitting based on the random sampling consistency principle and iterative reweighted fitting based on residual thresholds. By first filtering out outliers using a robust fitting strategy and then obtaining the plane normal vector using the interior point set, the anti-interference capability of the plane estimation can be significantly improved.

[0072] This step can still obtain reliable plane normal vectors even in the presence of local outliers, reducing the estimation error of the projection plane and thus improving the adaptability of the automatic trapezoidal correction model to complex production environments and different projection plane conditions.

[0073] The plane normal vector is transformed into a coordinate system established by the inertial measurement unit, and the yaw angle is calculated based on the plane normal vector and the horizontal reference direction of the coordinate system.

[0074] In this step, it should be noted that the set of spatial points output by the ranging sensor is usually located in the coordinate system of the ranging sensor itself or the coordinate system of the projector structure, while the attitude information is generally referenced to the coordinate system established by the inertial measurement unit.

[0075] Without coordinate unification, rotational deviations between different coordinate systems can lead to coupling errors between the projection plane parameters and attitude angles in the geometric model, thus affecting the accuracy of the trapezoidal correction transformation parameters. Therefore, this step transforms the plane normal vector to the inertial measurement unit coordinate system, ensuring that the projection plane direction and attitude parameters are within the same reference frame. Subsequently, the yaw angle is calculated based on the plane normal vector and the horizontal reference direction of this coordinate system. The yaw angle characterizes the left-right rotational deviation of the projector relative to the projection plane in the horizontal plane.

[0076] This step unifies the coordinate system and calculates the yaw angle, enabling the subsequent projected geometric model to accurately calculate using the roll angle, pitch angle, and yaw angle simultaneously. This reduces error coupling between angular quantities, improves the prediction accuracy of the initial trapezoidal correction model, and makes the subsequent calibration error more concentrated in reflecting the differences in single-machine assembly, thereby improving the stability and effectiveness of the compensation parameters.

[0077] The steps of establishing the projection geometric model and generating transformation parameters include: S21: Based on the optical emission parameters, determine multiple projection boundary rays corresponding to the boundary of the display panel.

[0078] In this step, it should be noted that the display panel boundary typically corresponds to the boundary pixel area of ​​the projected image, such as the upper, lower, left, and right boundaries of the display panel or their corner positions. The optical emission parameters are used to describe the emission geometry of the optical engine, preferably including the distribution of the emitted light direction of the optical engine, the mapping relationship from the display panel coordinates to the emitted light direction, and the emission vector corresponding to the projection boundary pixel point, etc.

[0079] By determining multiple projection boundary rays corresponding to the boundary of the display panel based on optical emission parameters, the boundary features of the display panel can be converted into a set of spatial rays with clear physical meaning, thus providing a basis for subsequent calculation of the true boundary of the projection area on the projection surface.

[0080] This step eliminates the need to rely on a camera to infer the boundary position from the projected image. The projected boundary can be directly calculated from the optical model, reducing the impact of image noise, ambient light, or projection surface texture on the boundary estimation. This makes the initial geometric model more stable and more suitable for rapid calibration on the production line.

[0081] S22: Calculate the intersection point of the projection boundary ray and the projection plane.

[0082] In this step, it should be noted that the projection plane is usually obtained by fitting a range-measuring lattice, and can be described by a plane equation or a plane normal vector and a point on the plane. Projection boundary rays can be represented by their origin, such as the projection optical center or an equivalent exit point, and a combination of direction vectors. Calculating the intersection points of projection boundary rays and the projection plane essentially involves solving for the intersection position of the ray and the plane in three-dimensional space, thus obtaining the actual spatial coordinates of the projection boundary on the projection plane. By finding the intersections of multiple boundary rays separately, a set of spatial intersection points covering the boundary of the projection area can be obtained.

[0083] This step can accurately map the projection area from the display panel coordinate system to the projection surface coordinate system, obtaining boundary points with actual size and orientation significance. This avoids boundary drift caused by rough estimation based solely on distance or attitude, thereby improving the accuracy of subsequent target correction area determination.

[0084] S23: The intersection point is determined as the set of boundary feature points of the projection area, and the target correction area is determined accordingly.

[0085] In this step, it should be noted that the intersection points of multiple projection boundary rays and the projection plane can be regarded as a set of sampling points of the projection area boundary. By organizing these intersection points into a set of boundary feature points, the boundary shape of the projection area on the projection plane can be further obtained, preferably a trapezoidal boundary or an approximately trapezoidal boundary.

[0086] Determining the target correction area based on the set of boundary feature points is to select a target area from the actual projection area that meets the requirements of the preset display ratio, cropping strategy or the maximum usable rectangle, so that the trapezoidal correction image can maximize the use of the projection area while avoiding excessive edge stretching that would lead to a decrease in clarity or exceed the projection surface range.

[0087] The beneficial effect of this step is that by using the set of boundary feature points, the "correctable region" is transformed from an abstract concept into a computable geometric object, which provides a clear target region constraint when solving transformation parameters in the future. This improves the controllability and consistency of the trapezoidal correction results, and is especially suitable for ultra-short throw projections to maintain stable image output under different wall distances and placement postures.

[0088] Determining the target correction area includes the following steps: S24: Determine the trapezoidal boundary using the boundary feature points of the projection area.

[0089] In this step, it should be noted that the boundary feature points of the projection area usually originate from the intersection of the projection boundary rays and the projection plane. These points, in the projection plane coordinate system, reflect the true geometric position of the edge of the projected image. Since the projector experiences roll, pitch, or yaw deviations, the shape of the projected image on the projection plane is generally trapezoidal or approximately trapezoidal. Therefore, organizing and fitting the boundary feature points to obtain a trapezoidal boundary, such as four boundary lines or four corner points, can clarify the geometric outline of the projection area, serving as a constraint condition for subsequently selecting the target correction area.

[0090] This step parameterizes the actual shape of the projection area as a trapezoidal boundary, giving the target area selection a unified geometric basis. This avoids problems such as image overflow, excessive edge stretching, or inconsistent selection areas on different devices caused by cropping based solely on experience, thereby improving mass production consistency and the stability of subsequent transformation parameter solutions.

[0091] S25: Find the inscribed rectangle with the largest area within the trapezoidal boundary, and use the vertices of the inscribed rectangle as feature points of the target correction region.

[0092] In this step, it's important to note that the trapezoidal boundary represents the actual projection area that the projector can cover in its current orientation. Trapezoidal correction essentially maps the projected image to a regular rectangular display through geometric transformation. Directly using the trapezoidal boundary as the target area for mapping can easily lead to problems such as the mapped rectangular image exceeding the trapezoidal coverage area, black borders, irregular cropping, or reduced edge sharpness.

[0093] Therefore, this step involves finding the inscribed rectangle with the largest area within the trapezoidal boundary. This ensures that the target correction area meets the requirements for rectangular display while maximizing the utilization of the current projection range, thereby maximizing the effective display area while maintaining image integrity. Using the vertices of this inscribed rectangle as feature points of the target correction area provides explicit corresponding point constraints for solving the homography transformation matrix, making the transformation parameter solution deterministic.

[0094] This step automatically selects the optimal display area without relying on manual adjustments, reducing black borders and cropping, improving screen utilization and visual consistency, and ensuring that different devices achieve consistent target area selection results under the same orientation, thereby improving mass production consistency and user experience.

[0095] The process of generating transformation parameters includes the following steps: S26: Establish the correspondence between the reference feature points in the display panel coordinate system and the feature points of the target correction area.

[0096] In this step, it should be noted that the reference feature points in the display panel coordinate system are usually preset standard points or screen corner points. These points serve as target positions to define the ideal target area to which the projector image should be corrected. For example, the reference feature points can be the coordinates of the four corners of the display panel, or some common reference positions set according to the aspect ratio and resolution of the display image.

[0097] The feature points of the target correction area are the vertices of the boundary or inscribed rectangle of the actual correction area on the defined projection surface. In this step, the correspondence between the display panel coordinate system and the actual projection area is established by comparing and mapping the physical positions of these points.

[0098] This step ensures that the calculation of subsequent transformation parameters has practical geometric meaning and consistency by clearly defining the correspondence between reference feature points and the actual correction area. This allows the image correction operation to be applied accurately to each projector, guaranteeing a consistent display effect.

[0099] S27: Solve for the homography transformation matrix based on the aforementioned correspondence.

[0100] In this step, it should be noted that the homography transformation matrix is ​​a geometric transformation matrix that describes the planar perspective transformation relationship between two sets of coordinate points, capable of mapping a set of reference feature points to a set of target feature points. The homography transformation matrix can be obtained by solving for the correspondence between the reference feature points and the feature points of the target correction area in the display panel coordinate system. The solution process typically employs the least squares method, deriving the transformation matrix through matched point pairs, with at least four corresponding points, thereby ensuring that the image can be accurately mapped from the projection plane to the display panel during subsequent correction.

[0101] The homography transformation matrix in this step provides an efficient and accurate geometric transformation method, which allows the mapping relationship between the corrected projected image and the display panel image to be precisely controlled through a mathematical model. This avoids image distortion or misalignment caused by directly using lookup tables or other coarse algorithms, ensuring the accuracy and operability of trapezoidal correction.

[0102] S28: Output geometric transformation parameters of the display panel image based on the homography transformation matrix, so as to serve as control parameters for the trapezoidal correction action.

[0103] In this step, it's important to note that the transformation parameters obtained after calculating the homography transformation matrix will be used to control the projector to perform trapezoidal correction. These geometric transformation parameters include information on the horizontal and vertical stretching, rotation, and shearing of the projected image, enabling the projected image to be transformed from its original projection state into a rectangular area consistent with the display panel's coordinate system. By applying these transformation parameters, the projector can accurately adjust the image distortion, thereby eliminating trapezoidal distortion caused by the non-perpendicular angle between the projector and the projection surface.

[0104] The control parameters output by the transformation matrix in this step make the trapezoidal correction process automated and precise. It completes the geometric correction of the image through mathematical calculations rather than human experience, reduces human intervention errors, improves the efficiency and stability of the correction operation, ensures that the trapezoidal correction effect is consistent under different devices and usage scenarios, and enhances the user experience.

[0105] The process of generating compensation parameters based on calibration error includes the following steps: S61: Obtain the calibration error and calculate the pitch angle compensation value under at least one positive projection attitude.

[0106] In this step, it's important to note that "front projection" refers to the projector's projection direction being perpendicular to the projection surface, meaning the projector is in a standard upright position and the projected image is rectangular. By obtaining calibration errors in this front projection position, such as the offset and distortion of the projection area boundaries, the required pitch angle compensation value relative to the ideal projection area in this position can be calculated. The pitch angle compensation value is used to correct the geometric distortion in the vertical direction of the image caused by assembly errors, optical path deviations, and other factors during actual projection.

[0107] This step can accurately calculate the pitch angle error of a single device in the forward projection posture, and provide a preliminary compensation benchmark for subsequent calibration in more complex postures, ensuring that the projected image of the projector remains stable and consistent in different postures.

[0108] S62: Obtain the calibration error under multiple different rotation angle attitudes and calculate the corresponding yaw angle compensation amount.

[0109] In this step, it's important to note that to accommodate the trapezoidal correction requirements of the projector at different placement angles, calibration errors need to be obtained and the corresponding yaw angle compensation calculated at multiple rotation angles, such as 2°, 3°, 4°, and 5°. Yaw angle compensation refers to the projector's rotational error relative to the projection surface in the horizontal direction, typically manifesting as left-right image distortion and curved image edges. By analyzing the calibration errors at different angles, the compensation amount for each angle can be accurately determined and further used to compensate for the projector's attitude errors in different usage scenarios.

[0110] This step calculates the calibration error at different rotation angles, enabling compensation for each specific angle. This allows the projector to achieve high-precision automatic keystone correction under various placement conditions, avoiding the shortcomings of single-angle compensation.

[0111] S63: Construct a fitting dataset from the rotation angle and the yaw angle compensation amount, and solve the compensation function coefficients for the left turn and right turn conditions respectively.

[0112] In this step, it's important to note that multiple different rotation angles, such as 2°, 3°, 4°, and 5°, along with their corresponding yaw angle compensation values, are combined into a dataset to facilitate subsequent function fitting. Because the projector's structure and optical design may lead to asymmetric compensation requirements at different rotation angles—for example, the errors differ when turning left and right—compensation functions need to be fitted separately for the "left turn" and "right turn" scenarios to obtain compensation coefficients for different rotation directions. Typically, the fitting method can employ least squares or other optimization algorithms to ensure that the fitted function accurately describes the variation of the compensation value during left and right turns.

[0113] This step, by fitting left-turn and right-turn conditions separately, can generate an independent compensation function for each condition, ensuring high-precision trapezoidal correction at any rotation angle and avoiding error accumulation caused by ignoring directional differences in traditional methods.

[0114] S64: During production line calibration, the pitch angle compensation value is adjusted based on the calibration error of the projector to be calibrated, and the constant term of the left turn compensation function or the right turn compensation function is corrected and then written into the projector.

[0115] In this step, it's important to note that during production line calibration, for each projector to be calibrated, the calibration error is measured and the pitch angle compensation value is calculated, allowing for fine-tuning of the pitch angle. For left or right turn compensation functions, the yaw angle compensation is further adjusted by modifying the constant term of the compensation function. Thus, although the specific assembly tolerances and hardware characteristics of each projector may differ, adjusting the constant term of the compensation function can eliminate the impact of these differences on the keystone correction accuracy, achieving a uniform compensation effect in mass production. This method not only ensures high precision in the projector's keystone correction but also avoids the need for complete full-scale calibration of each device during production, saving time and costs.

[0116] This step, by fine-tuning the pitch angle and the constant term of the compensation function, enables each projector to obtain personalized compensation parameters during mass production, thereby ensuring that the trapezoidal correction effect of each device meets the predetermined standard and improving production efficiency and consistency.

[0117] In one specific embodiment, to improve the consistency and calibration efficiency of mass production, the calibration process can be repeatedly executed on multiple prototypes at multiple preset rotation angles to obtain the pitch angle compensation value and the coefficient of the yaw angle compensation function for each prototype. Then, the above compensation data is statistically processed, for example, by taking the mean or median, to obtain a set of general compensation parameters, which are then written into the model as default production parameters. In this way, during production line calibration, each projector to be calibrated does not need to re-fit all compensation functions completely; it only needs to adjust the pitch angle compensation value in the forward projection posture and correct the constant term of the corresponding yaw angle compensation function in the left or right projection posture. This can absorb the assembly differences of individual machines and meet the preset geometric accuracy requirements, thereby balancing calibration accuracy and production line cycle time.

[0118] An automatic keystone correction system for a projector is provided, applicable to any of the automatic keystone correction methods described above. The system integrates multiple hardware and software modules, including the projector body, inertial measurement unit (IMU), distance sensor, image acquisition device, computing and processing unit, memory, and user interface and control unit. It automatically acquires projector attitude information and spatial information, calculates calibration errors, generates compensation parameters, and writes them into the projector to ensure that the projector can stably perform high-precision keystone correction under different usage scenarios.

[0119] The projector itself is responsible for performing trapezoidal correction, including projecting a preset calibration pattern and executing trapezoidal correction actions calculated based on compensation parameters. The IMU (Integrated Measurement Unit) acquires the projector's attitude information, including roll and pitch angles, providing accurate data on projector angle changes. A ranging sensor acquires spatial information of the projection surface, providing the geometric relationship between the projection surface and the projector through dot matrix ranging data. An image acquisition device acquires the calibration pattern projected by the projector and extracts image features, such as checkerboard corner points, obtaining the geometric parameters of the actual projected image through image processing algorithms. These data are processed and analyzed by a computational processing unit to calculate geometric parameters related to the projector's current state, generate transformation parameters, and compare them with the actual calibration results to calculate the calibration error and generate corresponding compensation parameters. The computational processing unit is also responsible for adjusting the pitch and yaw compensation values. It generates compensation functions by fitting datasets and fits the compensation function coefficients for left and right turns, ensuring that each device receives personalized compensation parameters during calibration. On the production line, only fine-tuning of the pitch compensation value and the constant term of the compensation function is required, achieving high-precision, low-cost calibration. The memory stores the generated compensation parameters and provides real-time correction during subsequent keystone correction, ensuring the projected image meets predetermined standards. The user interface and control unit provide an interactive interface, allowing operators to set and monitor the calibration process, controlling its execution and ensuring the projector's correction effect under various operating conditions. This system improves the projector's correction accuracy, consistency, and stability in production through automation, reduces manual intervention, increases production efficiency, and significantly enhances the user experience.

[0120] An automatic keystone correction medium for projectors is provided, applicable to any of the automatic keystone correction methods described above. Further details will not be elaborated upon here.

[0121] Therefore, the aforementioned automatic keystone correction method for projectors does not consider the process complete after the system and medium undergo preliminary correction using a geometric model. Instead, it actively projects a known calibration pattern and acquires the actual geometric parameters of the corrected image through image acquisition. Subsequently, the solution converts this measured error into compensation parameters specific to that particular device and permanently writes them into its memory. Thus, whenever the user performs keystone correction in any future usage scenario, the system will call upon these individual compensation parameters to correct the general model calculation results in real time.

[0122] The above description is merely an embodiment of this application. It should be noted that those skilled in the art can make improvements without departing from the inventive concept of this application, but these improvements all fall within the protection scope of this application.

Claims

1. An automatic keystone correction method for a projector, applied to ultra-short-throw projectors of the same model in production, characterized in that, The method includes the following steps: Obtain the attitude information of the projector and the spatial information of the projector and the projection surface; Given the optical emission parameters of the projector, a projection geometry model is established based on the attitude information and the spatial information, and transformation parameters for trapezoidal correction of the projector are generated accordingly. Based on the transformation parameters, the projector is controlled to perform a trapezoidal correction action; The projector is controlled to project a preset calibration pattern. After the trapezoidal correction action is performed, an image acquisition device is used to acquire an image of the calibration pattern and extract image features. Based on the image features, geometric parameters corresponding to the actual projected image are obtained. The calibration error is determined based on the difference between the geometric parameters and the transformation parameters, and compensation parameters are generated based on the calibration error. The compensation parameters are written into the projector to compensate for the transformation parameters when the projector performs keystone correction.

2. The automatic keystone correction method for a projector according to claim 1, characterized in that, Obtaining attitude information includes the following steps: Place the projector on a horizontal reference plane; Acquire the triaxial acceleration sampling sequence of the inertial measurement unit within a preset sampling time; The average value of the three-axis acceleration sampling sequence is calculated to obtain the three-axis acceleration bias. In subsequent attitude calculations, the corrected acceleration value is obtained by subtracting the triaxial acceleration offset from the real-time triaxial acceleration value, and the roll and pitch angles are calculated based on the corrected acceleration value.

3. The automatic keystone correction method for a projector according to claim 2, characterized in that, Obtaining spatial information includes the following steps: Acquire dot matrix ranging data output by the ranging sensor; The dot matrix ranging data is converted into a spatial point set based on the field of view parameters of the ranging sensor; Points that exceed a preset distance range or whose distance measurement is invalid are removed from the spatial point set; Fit the projection plane to the remaining point set to obtain the plane parameters.

4. The automatic keystone correction method for a projector according to claim 3, characterized in that, Fitting the projection plane includes the following steps: Select no less than a preset number of valid points from the set of spatial points; A robust plane fitting strategy is used to screen out outliers and obtain the plane normal vector; The plane normal vector is transformed into a coordinate system established by the inertial measurement unit, and the yaw angle is calculated based on the plane normal vector and the horizontal reference direction of the coordinate system.

5. The automatic keystone correction method for a projector according to claim 1, characterized in that, The steps to establish a projection geometry model and generate transformation parameters are as follows: Based on the optical emission parameters, multiple projection boundary rays corresponding to the boundary of the display panel are determined; Calculate the intersection point of the projection boundary ray and the projection plane; The intersection points are determined as the set of boundary feature points of the projection area, and the target correction area is determined accordingly.

6. The automatic keystone correction method for a projector according to claim 5, characterized in that, Determining the target correction area includes the following steps: The trapezoidal boundary is determined by the boundary feature points of the projection area; Find the inscribed rectangle with the largest area within the trapezoidal boundary, and use the vertices of the inscribed rectangle as feature points of the target correction region.

7. The automatic keystone correction method for a projector according to claim 5, characterized in that, Generating transformation parameters includes the following steps: Establish the correspondence between the reference feature points in the display panel coordinate system and the feature points of the target correction area; Solve for the homography transformation matrix based on the aforementioned correspondence; The geometric transformation parameters of the display panel image are output based on the homography transformation matrix and used as control parameters for the trapezoidal correction action.

8. The automatic keystone correction method for a projector according to claim 1, characterized in that, The steps involved in generating compensation parameters based on calibration error are as follows: The calibration error is obtained and the pitch angle compensation value is calculated under at least one positive projection attitude; The calibration error was obtained under multiple different rotation angle attitudes, and the corresponding yaw angle compensation was calculated. The rotation angle and the yaw angle compensation amount are constructed into a fitting dataset, and the compensation function coefficients are solved for the left turn and right turn conditions respectively. During production line calibration, the pitch angle compensation value is adjusted based on the calibration error of the projector to be calibrated, and the constant term of the left turn compensation function or right turn compensation function is corrected and then written into the projector.

9. An automatic keystone correction system for a projector, characterized in that, The automatic keystone correction method for projectors as described in any one of claims 1-8 above.

10. An automatic keystone correction medium for a projector, characterized in that, The automatic keystone correction method for projectors as described in any one of claims 1-8 above.

Images