CHAPTER 1 INTRODUCTION

Transcription

1 CHAPTER 1 INTRODUCTION Digital Image Processing deals with the acquisition, filtering, edge detection, segmentation, interpretation and identification of objects in an input image. In 1970s and onwards Digital Image Processing Proliferated, when computers and dedicated hardware became available. Over the past 35 years, there has been much interest in the automatic processing and analysis of digital images, and many valuable techniques have been developed. Much of the running has been made by the need of rapidly processed the enormous quantities of image data obtained from satellite, though medical and commercial applications have been important. Simultaneous with these activities have been efforts to understand and emulate the workings of Human Visual System, and this had led to the subject of computer vision. However, computer vision does not aim to understand biological vision in detail, but rather to build up a knowledge of what is involved in seeing, by finding what computational constructs are required if visual perception is to occur. With the knowledge of the possible computational constructs, neurophysiologists will be better equipped to unravel the workings of the eye-brain system, and many efforts have been made in this direction. Machine vision is distinct from computer vision in that it aims to make machines process images from the real world, thereby enabling them to perform certain necessary tasks-i.e., it is task oriented rather than understanding oriented. Machine vision tends to be oriented to the solution of specific tasks in specific environment. In fact, such tasks can be exacting and complex: they should not be taken as parochial or trivial. In manufacturing environments, there are a number of general functions to be performed by a machine vision system. These include: 1

2 Chapter 1: Introduction Control of robots performing assembly operations (choose for pick and place) Guidance of lasers during cutting, milling or welding operations (measure, plan and perform) Inspection of products during manufacture (check, select/reject) Feedback to control manufacturing processes(check, integrate data and suggest) General process monitoring Guidance of vehicles in factory And many more. Broadly these class can be classified into two main categories - guidance and inspection, which respectively involve mainly active and passive observations of manufacturing processes. To further distinguish the two categories, one can say that if an active observation process helps in smooth running and continuity of the process time, whereas passive operation look for maintaining the desired quality. Image processing and machine vision are likely to score high for quality control purposes. However, the principles and technology of machine vision for quality control are almost identical to those required for controlling robots and for guiding robot vehicles. This means that it will be possible to use virtually identical techniques for a great many purposes of a current relevance in the food industry and agriculture. Such purposes include: Guiding fruit picking machines Guiding pruning machines Guiding crop spraying machines Tracking and sizing animals Checking products for size and shape Checking icing patterns on cakes Inspecting products for appearance 2 I

3 Chapter 1: Introduction Analyzing quality of food products Controlling packing machines and many similar cases. These fall into general categories of inspection, handling, control and guidance. Computer vision is the construction of explicit and meaningful descriptions of physical objects from images (Ballard and Brown, 1982). It (Timmermans, 1998) encloses the capturing, processing and analysis of human vision by electronically perceiving and understanding an image (Sonka et at, 1999). Image processing and image analysis are the core of computer vision with numerous algorithms and methods available to achieve the required classification and measurements (Krutz et at, 2000). Computer vision systems have been used increasingly in the food and agricultural industry for inspection and evaluation purposes as they provide suitably rapid, economic, consistent and objective assessment (Sun, 2000). They have proved to be successful for the objective measurement and assessment of several agricultural products (Timmermans, 1998). Over the past decade, advances in hardware and software for digital image processing have motivated several studies on the development of these systems to evaluate the quality of diverse and processed foods (Locht et at, 1997; Gerrard et at, 1996). Computer vision has long been recognized as a potential technique for the guidance or control of agricultural and food processes (Tillett, 1990). Therefore, over the past 20 years, extensive studies have been carried out, thus generating many publications. The majority of these studies focused on the application of computer vision to product quality inspection and grading. Traditionally, quality inspection of agricultural and food products has been performed by manual grading. However, in most cases these manual inspections are time-consuming and labor-intensive. Moreover the accuracy of the tests cannot be guaranteed (Park et at, 1996). By contrast it has been found that computer vision 3

4 Chapter!: Introduction inspection of food products was more consistent, efficient and cost effective (Lu et at, 2000; Tao et at, 1995a). Also with the advantages of superior speed and accuracy, computer vision has attracted a significant amount of research aimed at replacing human inspection. Recent research has highlighted the possible application of vision systems in other areas of agriculture, including the analysis of animal behavior (Sergeant et at, 1998), applications in the implementation of precision farming and machine guidance (Tlllett and Hague, 1999), forestry (Krutz et at, 2000) and plant feature measurement and growth analysis (Wrren, 1997). Besides the progress in research, there is increasing evidence of computer vision systems being adopted at commercial level. This is indicated by the sales of Application Specific Machine Vision (ASMV) systems into the North American food market, which reached 65 million dollars in 1995 (Locht et at, 1997 and Gunasekaran, 1996) reported that the food industry is now ranked among the top ten industries using machine vision technology. A computer vision system generally consists of five basic components: illumination, a camera, an image capture board (frame grabber or digitizer), computer hardware and software as shown in Fig 1.1. (Wang & Sun, 2002a). Fig 1.1: Components of a computer vision system (Wang & Sun, 2002a). As with the human eye, vision systems are affected by the level and quality of illumination. In agreement (Gunasekaran, 1996) noted that a welldesigned illumination system can help to improve the success of the image 4

5 Chapter 1: Intivduction analysis by enhancing image contrast. Good lighting conditions can reduce reflection, shadow and some noise giving decreased processing time. Various aspects of illumination including location, lamp type and colour quality, need to be considered when designing an illumination system for applications in the food.industry (Bachelor, 1985). Most lighting arrangements can be grouped as either front or back lighting (Gunasekaran, 2001). Front lighting (electron projection lithography or reflective illumination) is used in situations where surface feature extraction is required such as defect detection in apples (Yang, 1994). In contrast back lighting (transmitted illumination) is employed for the production of a silhouette image for critical edge dimensioning or for subsurface feature analysis as in the size inspection of chicken pieces (Soborski, 1995). Light sources also differ but may include incandescent, fluorescent, lasers, X-Ray tubes and infrared lamps. The choice of lamp affects quality and image analysis performance (Bachelor, 1985). The elimination of natural light effects from the image collection process is considered of importance with most modem systems having built in compensatoiy circuitry. There are many different sensors, which can be used to generate an image, such as ultrasound, X-Ray and near infrared spectroscopy. Images can be also obtained using displacement devices and documents scanners. Typically the image sensors used in machine vision are usually based on solid state charged coupled device (CCD) camera technology with some applications using thermionic tube devices. The CCD cameras are either of the array type or line scan type. Arrays or area, type cameras consist of a matrix of photosensitive elements (photosites) from which the complete image of the object is obtained based on output proportional to the amount of incident light. Alternatively, line san cameras use a single line of photosites, which are repeatedly scanned up to 2000 times per minute to provide an accurate image of the object as it moves under the sensor (Wallin & Haycock, 1998). Monochrome and colour cameras have been used throughout the food industry 5

6 Chapter 1: Introduction for a variety of applications (Leemans, " - si*:, 1998 ; Pearson 8s Slaughter, 1996; Steinmetz et at, 1999; Yang, 1996). The X-ray radiography has also been used for the generation of images for computer vision analysis of a variety products such as water core in apples (Kim & Schatzki, 2000) and for the detection of bones in chicken and fish (Jamieson, 2002). Table 1.1 shows the different applications using X-ray imaging in computer vision. Table 11: Applications using X-ray Imaging in Machine Vision Application Accuracy Reference (%) Detection of bones m 99 Jamieson (2002) Fish and chicken Internal defects of sweet onions 90 Tollner, Shahin, Maw, Gitaitis, and Summer (1999) Spit pits m peaches 98 Han Bowers, and Dodd Water core damage in apples 92 Kim and Schatzki (2000) Pinhole damage in almonds 00 Kim, and Schatzki (2001) The process of converting pictorial images into numerical form is called digitization. In this process, an image is divided into a two dimensional grid of small regions containing picture elements defined as pixels by using a vision processor board called a digitser or frame grabber. There are numerous types of analog to digital converters (ADC) but for real time analyses, a special type is required, this is known as a flash ADC. Such flash devices require only nanoseconds to produce a result with megasamples processed per second (Davies, 1997). Selection of the frame grabber is based on the camera 6

7 Chapter 1: Introduction output, spatial and grey level resolutions required, and the processing capability of the processor board itself (Gunasekaran & Ding, 1993). Image processing and image analysis are recognized as being the core of computer vision (Krutz, Gibson, Cassens, & Zhang, 2000). Image processing involves a series of image operations that enhance the quality of an image in order to remove defects such as geometric distortion, improper focus, repetitive noise, non-uniform lighting and camera motion. Image analysis is the process of distinguishing the objects (regions of interest) from the background and producing quantitative information, which is used in the subsequent control systems for decision making. Image processing/analysis involves a series of steps, which can broadly divided into three levels: low level processing, intermediate level processing and high level processing (Gunasekaran & Ding, 1993 ; Sun, 2000), as indicated in Fig 1.2. (Sun, 2000). Low-level processing includes image acquisition and pre-processing. Image acquisition is the transfer of the electronic signal from the sensing device into a numeric form. Image pre-processing refers to the initial processing of the raw image data for correction of geometric distortions, removal of noise, gray level correction and correction for blurring (Shirai, 1987). Pre-processing aims to improve image quality by suppressing undesired distortions or by the enhancement of important features of interest. Averaging and Gaussian filters are often used for noise reduction with their operation causing a smoothing in the image but having the effect of blurring edges. Also through the use of different filters fitted to the CCD cameras images from particular spectral regions can be collected. Rigney, Brusewitz, and Kranzler (1992) used a nm interference filter to examine contrast between defect and good asparagus tissue. A multi-spectral camera system with six band pass filters for the inspection of poultry carcasses was used to achieve better classification of abnormal carcasses (Park & Chen, 1994). 7

8 Chapter J: Tntimhirtinn Fig 1.2: Different Levels in the Image Processing Process (Sun, 2000) Intermediate level processing involves image segmentation, and image representation and description. Image segmentation is one of the most important steps in the entire image processing technique, as subsequent extracted data are highly dependent on the accuracy of this operation. Its main aim is to divide an image into regions that have a strong correlation with object or areas of interest. Segmentation can be achieved by three different techniques: thresholding, edge-based segmentation and reigon-based segmentation. Thresholding is a simple and fast technique for characterizing image regions based on constant reflectivity or light absorption of their surfaces. Edge-based segmentation relies on edge detection by edge operators. Edge operators detect discontinuities in grey level, colour texture, etc. Region segmentation involves the grouping together of similar pixels to form regions representing single objects within the image. The criteria for like-pixels can be based on grey level, colour and texture. The segmented image may then be represented as a boundary or a region. Boundary representation is suitable for analysis of size and shape features while region representation is used in the 8

9 Chapter]: Tntmthwtion evaluation of image texture and defects. Image description (measurement) deals with the extraction of quantitative information from the preciously segmented image regions. Various algorithms are used for this process with morphological, texture, and photometric features quantified so that subsequent object recognition and classifications may be performed. High level processing involves recognition and interpretation, typically using statistical classifiers or multilayer neural networks of the region of interest. These steps provide the information necessary for the process/machine control for quality sorting and grading. The interaction with a knowledge database at all stages of the entire process is essential for more precise decision making and is seen as an integral part of the image processing process. The operation and effectiveness of intelligent decision-making is based on the provision of a complete knowledge base, which in machine vision is incorporated into the computer. Algorithms such as neural networks, fuzzy logic and genetic algorithms are some of the techniques of building knowledge bases into computer structures. Such algorithms involve image understanding and decision making capacities thus providing system control capabilities. Neural network and fuzzy logic operations have been implemented successfully with computer vision in the food industry (Ying, Jing, Tao, & Zhang, 2003). Computer vision systems are being used increasingly in the food industry for quality assurance purposes. The system offers the potential to automate manual grading practices thus standardizing techniques and eliminating tedious human inspection tasks. Computer vision has proven successful for the objective; online measurement of several food products with applications ranging from routine inspection to the complex vision guided robotic control (Gunasekaran, 1996). 9

10 Chapter 1: Introduction objectives: Thus the present research work was undertaken with following 1) Automatic quality assessment of different horticultural products using machine vision. 2) Automatic segmentation of on-tree fruits using multiple feature based image processing algorithm. 3) Automatic segmentation and Yield calculation of on-tree fruits using shape analysis. 4) Non-Destructive porosity calculation of Indian fermented food KHAMAN using X-ray Microtomography and image processing. 5) Non-Destructive quality analysis of Indian fermented food KHAMAN using image processing. 10