JPH04349496A - Device and system for image processing - Google Patents

Abstract

PURPOSE:To reduce circuits and the device in scale as to the image processor and its system for the enlargement and reduction of moving pictures and still pictures. CONSTITUTION:This processor and system are equipped with a frame memory 101 storing input image data, a buffer memory 102 storing display output data, a coordinate conversion part 103 which calculates original image coordinates from the address of the buffer memory 102 and outputs corresponding reference data to the frame memory 101, and a data processing part 104 which performs an interpolating process according to the reference data outputted from the frame memory 101 to generate data for each picture element and inputs the data to the buffer memory 102. When an input image is a still picture, the interpolating process is performed by bilinear interpolation or tertiary interpolation and when the image is a moving picture, the interpolating process is performed by the most neighbourhood method.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates primarily to an image processing apparatus and method for enlarging and reducing moving images and still images.

0002

[Prior Art] In recent years, many presentation images using natural images have been seen at various events and presentations, but these applications require real-time enlargement of videos, rotation, reduction of still images, etc. Processing functions are required, and many image processing devices having these functions have been commercialized. In this image processing device, coordinate transformation processing of an image is performed, and the coordinate transformation processing consists of the following two processes.

(1) Coordinate conversion processing that calculates the corresponding coordinates on the screen before processing from the coordinates on the screen after processing. (2) Interpolation processing that calculates the unprocessed coordinate data obtained by the above coordinate transformation (generally, there is no data because it is a non-lattice point) from surrounding data. In particular, the interpolation process (2) above is important because it affects the quality of the finished image, and a highly accurate interpolation method is required to generate a high-quality image. Image data interpolation methods include the "nearest neighbor method," which replaces the data of the nearest pixel with the processed data from the coordinates of the original image obtained by coordinate transformation, and the "nearest neighbor method," which replaces the data of the nearest pixel with the processed data, and the "nearest neighbor method," which replaces the data of the nearest pixel with processed data from the original image coordinates obtained by coordinate transformation. The "bilinear interpolation method" which performs interpolation based on 16 pixels, and the "cubic interpolation method" which performs interpolation based on 16 pixels are known.

FIG. 3 is a block diagram showing a conventional image processing device, in which numeral 11 is a frame memory for temporarily pooling input moving images or still images, numeral 12 is a buffer memory for display images, and numeral 13 is a coordinate transformation for performing coordinate transformation. 14 is a data processing section, and the coordinate transformation section 13 is a "bilinear interpolation method" or "
It is designed to generate coordinates ``address'' for performing ``cubic interpolation''.

[0005]

By the way, in conventional image processing apparatuses, the same complementary method is used for both moving image processing and still image processing. In order to improve the quality of the processed image, it is necessary to use ``bilinear interpolation'' or ``cubic interpolation,'' which require 4 pixels or 16 pixels, respectively, to generate data for 1 pixel. data is required. Therefore, when these interpolation methods are applied to video processing, the frame memory requires 4 or 16 accesses per dot clock (10-odd MHz), so in reality, 4 or 16 memories are required. A method is used to access the data in parallel. Therefore, not only is memory required for 4 or 16 screens, but data also needs to be generated for each dot clock, which increases the scale of the hardware that processes them, resulting in a very large device. There was a problem that I had no choice but to do.

[0006] The present invention was made in consideration of the above circumstances, and it is possible to reduce the size of circuits and devices by switching the data interpolation method associated with image processing according to each of moving images and still images. The purpose of the present invention is to provide an image processing device and its method that can perform measurement.

[0007]

[Means for Solving the Problem] As shown in FIG. 1, the present invention provides a frame memory 101 for storing input image data.
, a buffer memory 102 that stores display output data, a coordinate conversion unit 103 that calculates the original image coordinates from the address of the buffer memory 102 and outputs the corresponding reference data to the frame memory 101; a data processing unit 104 that performs interpolation processing based on reference data to generate data for each pixel and input it to the buffer memory 102; The present invention provides an image processing device that performs interpolation processing based on the nearest neighbor method when an input image is a moving image.

The present invention also provides an image processing method for enlarging/reducing moving images and still images, in which a data interpolation method accompanying image processing is switched corresponding to each of moving images and still images, and for moving images, The present invention provides an image processing method characterized in that processing is performed using an interpolation method in which a processed image has a lower quality than a still image. Note that it is preferable that the data interpolation process for moving images is performed based on the nearest neighbor method, and the data interpolation process for still images is performed based on a bilinear interpolation method or a cubic interpolation method.

[0009]

[Operation] In FIG. 1, the coordinate transformation unit 10 during video processing
3 generates the coordinates of the "nearest neighbor method" and stores the frame memory 1
Read data from 01. The read data is input to the buffer memory 102 and interpolated using the "nearest neighbor method", resulting in a display image of low quality. However, due to the characteristics of human vision, it is difficult to perceive details in moving images, so even images of a certain low quality can be viewed without causing any problems.

[0010] When processing a still image, the human senses can perceive considerable detail, so the coordinate transformation unit 103
The output data of the frame memory 101 is processed by the data processing unit 10.
4, the data is interpolated and input to the buffer memory 102. Since this still image processing operation is allowed to be performed at a certain low speed, no inconvenience occurs even if data is generated after accessing the frame memory 101 4 or 16 times.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below based on embodiments shown in the drawings. This invention is not limited by this. FIG. 2 is a block diagram showing one embodiment of the present invention. In FIG. 2, numeral 1 denotes a frame memory which is composed of a dual port memory and stores input image data as it is, and numeral 2 is a buffer memory which is composed of a dual port memory and stores display screen data. 5
is an image input unit that generates an input side write command and address for writing input image data into the frame memory 1; In addition, 3 is a coordinate conversion unit, and during video processing, display screen coordinates (address of buffer memory 2)
The coordinates of the original image are calculated using a coordinate transformation method known as affine transformation, and an address and a read command for retrieving reference data, that is, one piece of data necessary for the nearest neighbor method, from the frame memory 1 are generated. In addition, during still image processing, the coordinate transformation unit 3 calculates the original image coordinates from the display screen coordinates (addresses in the buffer memory 2) using an affine transformation method, and obtains reference data, that is, four-point data necessary for the bilinear interpolation method. Alternatively, an address and a read command for retrieving 16 points of data necessary for the cubic interpolation method from the frame memory 1 are generated.

Note that during still image processing, the degree of deviation of a non-lattice point from a lattice point that occurs during affine transformation is output from the coordinate transformation section 3 to the data processing section 4. Reference numeral 4 denotes a data processing unit that operates only when processing still images, and when reference data is output from the frame memory 1 to the data bus B according to the address and read command output by the coordinate conversion unit 3, the coordinate conversion unit 3 calculates the reference data. Interpolation is performed using the reference data and the deviation from the grid points to generate data for each pixel and output to data bus B.

Reference numeral 6 denotes a write control unit, which receives from the coordinate conversion unit 3 the output from the frame memory 1 to the data bus B by the coordinate conversion unit 3 during video processing, and
The address and write command for writing the data into the buffer memory 2 are output. Furthermore, when processing a still image, the write control unit 6 receives the data output from the data processing unit 4 onto the data bus B.
The address and write command for writing the data into the buffer memory 2 are output.

The operation in such a configuration will be explained next. First, when a switching signal instructs the coordinate conversion section 3 and the data processing section to perform video processing or still image processing, image processing is performed in response to the command. During moving image processing, the image input section 5 continues to write input image data into the frame memory 1 and constantly updates its contents. Coordinate conversion section 3
generates coordinates using the "nearest neighbor method" and reads data from the frame memory 1. The read data is transferred via data bus B and written into buffer memory 2 by write control section 6.

Furthermore, during still image processing, the image input section 5 stops writing input image data to the frame memory 1. The coordinate conversion section 3 generates coordinates using the "bilinear interpolation method" or "cubic interpolation method," and the output data of the frame memory 1 is inputted one byte at a time to the data processing section 4, and then interpolated and sent to the data bus B. is transferred to the write control unit 6.
is written into the buffer memory 2 by the command.

[0016]

[Effects of the Invention] According to the present invention, the nearest neighbor method is used when processing moving images, and the interpolation method, which is more accurate than the nearest neighbor method, is used when processing still images, so the number of frame memories is reduced and The data processing section can be simplified, and the processing circuit and device can be downsized.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the principle of the invention.

FIG. 2 is a block diagram showing an embodiment of the invention.

FIG. 3 is a block diagram showing a conventional example.

[Explanation of symbols]

1 Frame memory 2 Buffer memory 3 Coordinate conversion section 4 Data processing section 5 Image input section 6 Light control section B Data bus

Claims (3)

[Claims] Claims: 1. A frame memory (101) for storing input image data, a buffer memory (102) for storing display output data, and an original image coordinate calculated from the address of the buffer memory (102) and corresponding reference data. The coordinate transformation unit (10
3), and a data processing unit (104) that performs interpolation processing based on the reference data output from the frame memory (101) to generate data for each pixel and input it to the buffer memory (102). An image processing device characterized by performing interpolation processing based on a bilinear interpolation method or cubic interpolation method when an input image is a still image, and performing interpolation processing based on a nearest neighbor method when an input image is a moving image. Claim 2: In an image processing method for enlarging/reducing moving images and still images, the data interpolation method accompanying image processing is switched corresponding to each of the moving images and still images, and for moving images, the processed image An image processing method characterized by processing using an interpolation method that provides lower quality than still images. 3. The image processing method according to claim 2, wherein data interpolation processing for moving images is performed based on a nearest neighbor method, and data interpolation processing for still images is performed based on a bilinear interpolation method or a cubic interpolation method. .