JP2006313415A - Image processing apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing apparatus performing resolution conversion processing of a digitally-coded image at a higher rate and with higher quality. <P>SOLUTION: The apparatus comprises an input means inputting a main image signal and a thumbnail image signal relating to the main image signal; a feature detection means detecting the feature of the main image signal corresponding to the thumbnail image signal on the basis of the thumbnail image signal; and a resolution conversion means determining a resolution conversion method to be applied to the main image signal from a plurality of kinds of resolution conversion methods on the basis of the detection results from the detection means and performing the resolution conversion processing to the main image signal by means of the determined conversion method. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image processing apparatus, and more particularly to a resolution conversion process for an image signal.

  In recent years, digital cameras have rapidly spread and are used by many users. A digital camera encodes a photographed image by a compression encoding method called JPEG (Joint Photographic Experts Group), and stores it as digital data in a medium such as an HDD or a memory card. The user browses this image with a PC or the like and uses it by printing it with a printer.

  There is also a demand for browsing JPEG images taken with a digital camera on a television screen. In order to view digitally encoded images, a decoder is required. If this decoder is prepared in the television, the digital image stored in the HDD in the television or the memory card can be transferred to the television. A digital image can be viewed on a television at any time by inserting it into a slot.

  Digital images from digital cameras have been improved in definition, and it is considered that digital images will be improved in the future. However, since the resolution of a television is not so high, for example, when a digital image is displayed on a full screen, a resolution conversion process is essential.

Various methods are considered for resolution conversion processing of digital images, and there are roughly divided resolution conversion in the spatial domain and resolution conversion in the frequency domain. Typical resolution conversion methods in the spatial domain include the nearest neighbor method, the bilinear method, and the bicubic method. The nearest neighbor method is a method in which the pixel of the original image closest to the pixel position after conversion is used as a new pixel value. Since the amount of calculation is small, the processing speed is high, but the image quality is very poor. The bilinear method is a kind of linear interpolation, and is a method of linear interpolation using four points near the pixel value to be obtained. The image quality is improved compared to the two-arrest neighbor method, but the processing speed is low. The bicubic method is a kind of cubic interpolation, and interpolation is performed using 16 pixels near the pixel value to be obtained. Although the processing speed is very slow due to a large amount of calculation, the image quality is a resolution conversion method that is generally better than the above two methods. The resolution conversion method in the frequency domain is a method of performing resolution conversion by deleting a coefficient obtained by DCT conversion or inserting a zero value. When a digital image is displayed on a television, resolution conversion is performed using any of these methods (see, for example, Patent Document 1).
Japanese Patent Laid-Open No. 11-8838

  However, when the resolution conversion process is performed as described above, it is impossible to desire both high-speed processing and high image quality. Taking the resolution conversion method in the spatial domain as an example, the nearest neighbor method can be used for high-speed processing, so digital images can be displayed instantaneously on the TV, but the image quality is poor, It is hard to say that it can withstand browsing. Also, when using the bicubic method, etc., the resolution is converted to high image quality, so there is no sense of incongruity when browsing, but it takes time to display due to the large amount of processing, and the instantaneous responsiveness is impaired. End up.

  An object of the present invention is to solve the problems of the above-described method and provide an image processing apparatus capable of obtaining a high-definition image even when a resolution conversion process is performed.

  In order to solve the above object, in the present invention, an input means for inputting a main image signal and a thumbnail image signal related to the main image signal, and a main image signal corresponding to the thumbnail image signal based on the thumbnail image signal A feature detection unit for detecting a feature of the image, and a resolution conversion method to be applied to the main image signal from a plurality of types of resolution conversion methods based on a detection result of the feature detection unit, and the determined resolution conversion method And a resolution conversion means for performing resolution conversion processing on the main image signal.

  According to the present invention, it is possible to perform resolution conversion of a digitally encoded image at high speed and high image quality by efficiently grasping the characteristics of the digitally encoded image using a thumbnail image.

  Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to examples.

  FIG. 1 is a block diagram showing the overall configuration of a resolution conversion system for digitally encoded images according to the present invention. Arrows between modules in the system indicate the flow of data, thick arrows indicate the flow of main data handled by the system such as image data, and thin arrows indicate control signals and the like. Hereinafter, each module will be described.

  The resolution conversion system 100 for a digitally encoded image is a resolution conversion system that inputs an encoded image 120 and outputs it as an output image 121 after performing decoding processing and resolution conversion processing. The input method of the encoded image 120 is a memory card such as a CF (Compact Flash (registered trademark)) card for storing images taken by a digital camera, and the output destination is a display device such as a television.

  The input encoded image 120 is separated into a thumbnail image and a digital encoded image by the thumbnail information separation unit 101. Thereafter, the digitally encoded image is sent to the image decoding unit 105, and the thumbnail image is directly sent to the feature extraction unit 103, or if the thumbnail is encoded and needs to be decoded, after being sent to the thumbnail decoding unit 102 It is sent to the feature extraction unit 103. The thumbnail image and the digitally encoded image hold data in the configuration shown in FIG. FIG. 2 shows the configuration of a DCF (Design rule for Camera File system) basic file, and is a configuration diagram of a file system standard adopted by most digital cameras in taking a current digital image. The digital image is divided into a header portion and a main data portion. A thumbnail 201 is stored in the header portion, and a digital encoded image 202 is stored in the main data portion. The thumbnail 201 is allowed to be either compressed or uncompressed data, but is usually JPEG compressed in the same manner as the digitally encoded image 202. Therefore, the thumbnail decoding unit 102 and the image decoding unit 105 basically have the same configuration. The thumbnail information separation unit 101 performs an operation of separating the thumbnail 201 and the digital encoded image 202 in FIG.

  The feature extraction unit 103 analyzes the thumbnail image separated by the thumbnail information separation unit 103 and performs an operation of extracting the feature of the digitally encoded image. In the extraction method, an edge detection filter is applied to each small area obtained by dividing the thumbnail image into a plurality of areas, and the presence / absence of edges in the small area (high-frequency component amount) is calculated and indicated by a value. This value becomes the feature information 110 and is sent to the resolution conversion method determination unit 104. The resolution conversion method determination unit 104 determines the resolution conversion method using the feature information 110. The resolution conversion method is determined by using the value of the feature information 110. The bicubic method is used when the value is a certain value or more, the nearest neighbor method is used when the value is less than a certain value, and the middle is located between them. If so, the bilinear method is selected and this is used as the resolution conversion method for the digitally encoded image corresponding to the small area of the thumbnail. The method determined by the resolution conversion method determination unit is sent to the resolution conversion unit 106 as a method instruction signal 111.

  Here, advantages of using thumbnails for feature extraction will be described with reference to FIG. FIG. 4 is a diagram schematically showing the sizes of the thumbnail image 401 and the digitally encoded image 402. When the thumbnail image 401 conforms to the DCF standard, the size is fixed and 160 × 120 pixels. Further, it is assumed that the digitally encoded image 402 is an image photographed at 640 × 480 pixels. When the above feature extraction is executed in units of 8 × 8 pixels in the thumbnail image 401, a region of 32 × 32 pixels is shown on the actual digital encoded image 402. That is, by applying the edge detection filter of 8 × 8 pixels to the thumbnail image 401 once, it is possible to determine the characteristics of the 16 times region on the digital encoded image 402 at a time. In FIG. 4, the digital encoded image 402 has a size of 640 × 480 pixels, but since the current resolution of digital images is mainly larger than this, digital encoding corresponding to a small area on the thumbnail image 401 is performed. The size on the image 402 is even larger. Therefore, by using the thumbnail image 401 for feature extraction, it is possible to efficiently grasp the characteristics of the entire digital encoded image 402.

  The digitally encoded image is decoded by the image decoding unit 105 and then sent to the resolution conversion unit 106. In the resolution conversion unit 106, a resolution conversion method according to the method instruction signal 111 is applied to the decoded digitally encoded image, and an output image 121 is obtained.

  The above is the description of the overall configuration of the resolution conversion system for digitally encoded images according to the present invention. The processing flow when the resolution conversion is actually performed using this system will be described with reference to FIG.

  FIG. 3 is a flowchart according to the embodiment of the present invention. FIG. 3 shows a processing procedure when resolution conversion is performed by decoding a digital image stored in the file system configuration shown in FIG.

  First, in step S301, it is determined whether the thumbnail is encoded. If encoded, the thumbnail is decoded in step S302, and actual image data (raw data) is obtained. Since it is possible to use an unencoded (uncompressed) thumbnail image as it is, it is not necessary to send the thumbnail image to the thumbnail decoding unit 102 in FIG. 1, and the process proceeds to the next step.

  Next, in step S303, feature extraction of thumbnail images is performed. In feature extraction, as shown in FIG. 4, the thumbnail is divided into small areas, and an edge detection filter is applied to each area. Then, using the obtained feature value for each small area, a resolution conversion method is determined in step S304. The resolution conversion method determined in step S304 is applied in step S305 to the digitally encoded image area corresponding to the small area on the thumbnail to execute resolution conversion.

  In step S306, it is determined whether resolution conversion has been performed for all areas. If not, the process returns to S303 and the processing flow is repeated. When all the regions are completed, the resolution conversion of one digitally encoded image is completed.

  In this embodiment, the feature detection of the thumbnail image is performed by applying the edge detection filter after decoding the thumbnail. However, other feature extraction filters can be used. In this case, the threshold value for determining the resolution conversion method and the corresponding resolution conversion method are different. In addition, when a thumbnail image is encoded, feature extraction can be performed without performing a decoding operation. When the thumbnail image is encoded by JPEG, orthogonal transformation called DCT (Discrete Cosine Transform) is applied. Since this is an operation for separating the low frequency component and the high frequency component from the block to be applied, the presence or absence of the high frequency component of the block can be determined by determining the DCT coefficient obtained by DCT.

  In the present embodiment, the resolution conversion system for digitally encoded images has been described. However, the same effect can be obtained by performing the same operation for the resolution conversion method for digitally encoded images.

  According to the present embodiment, by extracting features of a digitally encoded image using a thumbnail image, it is possible to grasp the features of the digitally encoded image faster than analyzing itself, Resolution conversion can be performed at high speed with high image quality.

It is a block diagram showing the resolution conversion system of the digitally encoded image concerning embodiment of this invention. It is a figure which shows the structure of the digital encoding image utilized in embodiment of this invention. It is a flowchart which shows embodiment of this invention. In an embodiment of the invention, it is a figure showing a difference in size between a thumbnail image and a digitally encoded image.

Explanation of symbols

100 Digital Coded Image Resolution Conversion System 101 Thumbnail Information Separation Unit 102 Thumbnail Decoding Unit 103 Feature Extraction Unit 104 Resolution Conversion Method Determination Unit 105 Image Decoding Unit 106 Resolution Conversion Unit

Claims (1)

Input means for inputting a main image signal and a thumbnail image signal related to the main image signal;
Feature detection means for detecting a feature of the main image signal corresponding to the thumbnail image signal based on the thumbnail image signal;
Based on the detection result of the feature detection means, a resolution conversion method to be applied to the main image signal is determined from a plurality of types of resolution conversion methods, and resolution conversion is performed on the main image signal by the determined resolution conversion method. An image processing apparatus comprising: resolution conversion means for performing processing.

Images