The Focus Evaluation Method Research In Pushbroom Remote Sensing Camera

Remote sensing imaging technology plays an extremely role in the national economy and national security. And as the core of remote sensing imaging technology, the remote sensor has been focused researched. Nowadays, pushbroom remote sensing camera based on TDICCD increasingly shows its importance in aerospace remote sensing technology, which is one kind of remote sensors. Remote sensing camera sometimes will be out-of-focus because of the influence of space environment. In order to obtain high quality images, remote sensing camera needs effective auto-focusing technology. However, pushbroom remote sensing camera has a special way of imaging, which causes the scenes imaged by the camera at arbitrary time be different. Consequently, it is time to propose a focus evaluation method that is robust for images with varying scenes, then the method can be applied to the auto-focusing technology of pushbroom remote sensing camera.According to the above background, this paper firstly introduces the imaging theory of pushbroom remote sensing camera within TDICCD. And based on the analysis of the existing focus evaluation methods used in pushbroom remote sensing camera, the paper puts forward two focus evaluation methods that is robust in variable imaging conditions from wavelet packet domain and spatial domain respectively. One focus evaluation method is based on the energy spectrum in wavelet packet domain, the other focus evaluation method is based on the histogram of strong edge width.In the method based on the energy spectrum in wavelet packet domain, this part firstly introduces the wavelet packet technology has two features----fine frequency resolution and excellent time-frequency characteristic. And based on the principle that most arbitrary scenes theoretically have the same power spectra in frequency domain, we find that most arbitrary scenes also have the same energy spectra in wavelet packet domain. The algorithm firstly decomposes the image using wavelet packet technology, then selects the wavelet packet coefficients at different frequencies and at the specific direction which is perpendicular to the camera movement direction. Next we can get the energy at corresponding frequency from the wavelet packet coefficients and finally we can gain the sharpness by weighted summing up the different energies to guide the camera focusing. According to the experiments, we can draw the conclusion: when the camera has no image motion mismatching or has image motion mismatching but the defocus isn’t serious, this method can realize absolute focus check, which can be independent on scenes.In the method based on the histogram of strong edge width, this part shows that remote sensing images are full of line structure and the defocus can make edge diffuse, so we can use the diffusion of edge to represent the sharpness. The algorithm firstly employs the gradient operator to select the strong edges, which is perpendicular to the camera movement direction and then calculate the width of each strong edge. Next we can obtain a histogram of strong edge width and based on the histogram, we proposed a distance factor. Then we introduce the factor into the model, which could gain the average width of strong edges by the weighted mean method, and finally get the sharpness to guide the camera focusing. From the experiments, we can know:this method also can realize absolute focus check, which can be independent on scenes. And the focus curve has fine monotony and it’s robust for image motion mismatching. In addition, it also has a high correlation with subjective assessment. Therefore, this method is an effective focus evaluation method for pushbroom remote sensing camera.