Technique for normalizing intensity histograms of images when the approximate size of the target is known: Detection of feces on apples using fluorescence imaging

A challenge in machine vision is to develop algorithms for detecting a substance with an amorphous shape when measured responses of both the substance and the underlying target have similar characteristics. The challenge is exacerbated when responses for targets are highly variable both across and within discrete target units. An example of this problem is the detection of fecal contamination on apples. Feces on apples can be detected using differential fluorescence responses of contaminated and uncontaminated apple surfaces to UV excitation. However, responses of both feces and apples are due to the presence of chlorophyll-related compounds, and the response of apples varies within and between apples due to natural variation in the distribution of these compounds. We present a technique for normalizing the variability of intensity responses among targets based on a priori knowledge of the image dimensions and the approximate target size. Using this information, a linear equation is derived based on the approximate median intensities of the background and of the target. The median intensities are estimated by calculating a cumulative intensity histogram and using a priori estimates of the percentage of the area in the image occupied by the background and by a generic target. The image is scaled for uniform intensity power using this linear transformation. The benefits of using this technique are demonstrated using hyperspectral fluorescence responses to UV excitation of 48 Golden Delicious and 48 Red Delicious apples artificially contaminated with dilutions of cow feces. Results show that the uniform power transformation normalizes the intensity distributions of apple images and increases the contrast between contaminated and uncontaminated areas on apple surfaces; the coefficients of variation for the average intensities of uncontaminated apple surfaces at 668nm for Golden and Red Delicious apples were reduced from 39 and 55%, respectively, to 5% for both varieties.

[1]  R. Tauxe Emerging foodborne pathogens. , 2002, International journal of food microbiology.

[2]  Arthur R. Weeks Fundamentals of electronic image processing , 1996, SPIE/IEEE series on imaging science and engineering.

[3]  Y. R. Chen,et al.  HYPERSPECTRAL REFLECTANCE AND FLUORESCENCE IMAGING SYSTEM FOR FOOD QUALITY AND SAFETY , 2001 .

[4]  Alan M. Lefcourt,et al.  Detection of fecal contamination in apple calyx by multispectral laser-induced fluorescence , 2005 .

[5]  J. Morris,et al.  Emerging foodborne pathogens: Escherichia coli O157:H7 as a model of entry of a new pathogen into the food supply of the developed world. , 1996, Epidemiologic reviews.

[6]  L. McCaig,et al.  Food-related illness and death in the United States. , 1999, Emerging infectious diseases.

[7]  Moon S. Kim,et al.  Automated detection of fecal contamination of apples by multispectral laser-induced fluorescence imaging. , 2003, Applied optics.

[8]  Moon S. Kim,et al.  Optimal fluorescence excitation and emission bands for detection of fecal contamination. , 2003, Journal of food protection.

[9]  Moon S. Kim,et al.  A transportable fluorescence imagining system for detecting fecal contaminants , 2005 .

[10]  大泉 康 Foodborne Disease Handbook , 1995 .

[11]  Y. R. Chen,et al.  Multispectral detection of fecal contamination on apples based on hyperspectral imagery: Part II. Application of hyperspectral fluorescence imaging , 2002 .

[12]  Yud-Ren Chen,et al.  Machine vision technology for agricultural applications , 2002 .

[13]  P. McClure,et al.  Foodborne pathogens : hazards, risk analysis and control , 2002 .

[14]  Yud-Ren Chen,et al.  Detection of fecal contamination on apples with nanosecond-scale time-resolved imaging of laser-induced fluorescence. , 2005, Applied optics.